Never was the horrible state of this game's server response time or inherent latency more painfully obvious than in this current tier of savage raiding. I do not feel as though it's acceptable or perhaps even intended that "part of raiding" is the need to counteract server latency by being overly deliberate with every movement or action. For example, consider AS4 (or A4 as the same mechanic exists there). The balls that spawn do quite a bit of damage and simply "running through" a ball or even into a ball and then turning to run away from it won't work to explode it. You have to run to it, stand in it for a full second or two, and then let it explode before you move. Further, this latency issue is extremely obvious in all kinds of aspects of this game. For example, the BLM ability Aetherial Manipulation "counts" you as being at your starting location even after you've arrived at your target. It takes a full second for AM to "register" that you've arrived at your target. If you don't believe me, go test it by standing until an AoE is about to go off underneath you and then AM to a target that is out of the AoE. You will get hit by the AoE despite being far out of it on your screen.
I love this game. It has a lot of great content and there's a ton of things I really, really enjoy about it from the crafting system, to the way many of the rotations work, to the raid bosses. However, every attempt at getting friends of mine to come to FFXIV from other MMOs is always met with the same argument: that they can't get used to and deal with the "lag" in this game.
Are there any plans to address this? It seems as though the servers are only "polling" once every second or two and this is extremely problematic - especially for savage raiding.
Thanks
Serious Performance Issues (latency/server response time)
Printable View
-
08-31-2015, 01:29 PMSynoviusSerious Performance Issues (latency/server response time)
-
09-01-2015, 04:52 AMRaist
The servers poll more than 3 times a second. More often than not the problem is network latency issues across the internet and not server side. Get your route to your server checked thoroughly---find the IP your client uses with Resource Monitor (resmon from search/run), run a trace to that address and submit it to your ISP's tier3 support for investigation. Would be a good idea to show a trace to another good service as well for comparison (another online game or a video service, etc). It is usually best to do this through your ISP's online contact options---phone support is typically entry level and won't be able to deal with this, and it can be an exercise in futility sometimes to get it escalated to a higher tier.
Can also do some testing with some VPN's during their free trial period (some even have a long-term free use policy, though it may be restricted use or locations). Try different tunnel destinations (sometimes the ones closest to Montreal aren't the best choice). If you can improve your stability/response times through a VPN, it is pretty much proof-positive there are issues along the route your ISP is assigning you. Routing issues like that fall under your ISP's responsibility to address and not SE---Square-Enix does not hold influence over your ISP's routing policies. -
09-01-2015, 04:56 AMmaTyaR
Sadly, the location of where they have decided to implement the data centers have a routing to one of the most terrible network routing possible.
The SE servers itself have no issues with latency itself (199.91.189.25 - Gilgamesh, getting consistent numbers through trace routing/ping), but to reach to their servers, we're forced to go through Level3 node servers. This is where the serious lag comes in if you ping one of the routing hops on Level3 (4.69.141.1 - massive latency spikes from 26ms to 230+ms with no packet loss). Level3 seems have to signed private contracts to reroute bandwidth priority to Netflix during heavy peak times (rumors).
Second issue is the game concept itself using TCP. To have a precise location of where you and the enemy stand, you would use UDP (custom programming). First person games use UDP strictly, as players would want as close to 1:1 precision of what you see, and what the opponent sees. FFXIV cannot use UDP as it uses more data traffic to calculate your positioning, the monsters, and everyone/everything else you see on your screen itself. It also causes more strain on the SE servers to determine the specific location, and sending the data back to every client.
In summary, Level3 is to blame for the latency spikes, SE made a game concept of mechanics which would normally require positional precision.
P.S: HI LALA, :D!!
Edit: And to off load the amount of data, lag compensation would be used.
Edit 2: This will only get worse. It needs to be addressed ASAP or else it'll be as devastating as player congestion during official launch. -
09-01-2015, 11:09 AMSynovius
Thanks for the replies.
As to it not being server-side and it being with our connection, I'm afraid to report that this isn't an issue isolated to me or even 2 or 3 or 5 people. This is an issue spread across literally everyone I've ever raided with. This is through 100mbps connections, through using Battleping or WTFast or even through VPNing to locations close to the data center. This is an issue with FFXIV and not with our machines or our connections. For example, here are my stats:
i5 2500K CPU
GTX980 4GB GPU
8GB RipJaws DDR3 RAM
Intel SSD
100mbps down / ~20mbps up
This is an issue with the game and with the servers or the way the game handles collision detection/positioning. I can even make a video if you want of me running directly through the balls on AS4 to have them not explode. I've even done this on AS3 where I've accidentally side-stepped through the little ball that comes out in the last phase only to have it not explode (thankfully in this case).
All I can say is that were I to fire up a game like WoW (I don't play WoW anymore because I find it boring and old), the responsiveness of the game is extremely crisp and exact. Things are far more fluid and synchronized there. I have no idea why this is but it is.
As to E'P's point, that makes sense with what limited understanding I have of networking concepts. I wonder if WoW uses UDP? At the very least I know the positioning of their data centers makes for substantially better routing. Although I could have sworn my latency to Gilgamesh was around 100ms which isn't amazing but it's all that bad (for an FPS it would be noticeable but shouldn't really be noticeable for an MMO). By comparison, the latency I feel in FFXIV, were I to compare it to my days of competitive UT2K4, would be like I was playing Bombing Run at 300-400ms.
Also HI E'P'! -
09-01-2015, 11:29 AMRaistVPN'ing to the tunnel nearest the endpoint has been shown to not always improve things...the reason being you may still be hitting the congested networks in the same area you hit without the VPN. I myself have found more stable connections going through the midwest then back over (I'm on the East Coast). I've even seen one VPN that reported it's geolocation as Hungary respond more consistently than one to Montreal (though it was notably slower, but it didn't spike all over the place). I recently demonstrated this with my TunnelBear account on these forums.Quote:
Originally Posted by Synovius
It all depends on where your particular problems stem from. If there are issues with local segments--you can still have the same issues with a VPN, because even though you are now encrypted you are still physically on the same medium. If those problems are because of too many people physically on the line--bypassing the shaping filters with encryption is going to have a minimal impact. If the problem is in/around Level3's over-utilized exchange points for Montreal out of the North East, and the VPN is still going through those same exchanges---you will still have the same problems. BUT...if you are able to switch it up and get on to Cogent, TATA, TiNet/GTT...then things might improve. (Those URL's are links to their network maps...notice how they service the same regions? It means ISP's often have options on who they peer with.)
And that is why you see it happen to large groups at once...they are getting shunted through the same congested corridors. Ormuco currently lists peering with 5 major ISP's:
http://chart.googleapis.com/chart?ch...er&chco=346090
AS701 Verizon Business/UUnet
AS3356 Level 3 Communications, Inc.
AS3257 Tinet SpA
AS6453 TATA COMMUNICATIONS (AMERICA) INC
AS174 Cogent Communications
(source: Eidos/Ormuco ASN report for the Canadian subnet 199.91.189.0/24 from Hurricane Electric's BGP tool kit)
Notice those percentages in the pie chart. If any one of those is having issues, it is affecting potentially at least 15-28% of all traffic. If it is TWO ISP's having issues, you could be in the realm of 31-50% of all traffic being affected.
That is why your WoW experience doesn't quite cut it. Last I recall, they didn't have servers in Montreal...the addresses listed there for the Americas come back with geolocation data for New Jersey and California. -
09-01-2015, 01:31 PMSynovius
While this information is all very interesting and informative, it doesn't change the fact that the poor performance of this game related to latency/lag/response time/whatever-you-want-to-refer-to-it-as affects literally everyone I've spoken too. Just an hour spent on ANY WoW server - even one across the United States from you (I live in EST and often played on PST servers) - yield crispness and seemingly immediate response times where moving out of an AoE a the very last split second is perfectly fine, moving into an AoE buff buffs you literally the split second you arrive in it and a Warrior's Charge ability tracks you every point of the way from your source to destination - so much so that you can be stunned mid-charge at any point and you will freeze immediately at that location.
So whether it's better data center locations (or more of them), or working with ISPs for better routing or reduction in any throttling (if that is even occurring), Square Enix needs to do something and do something relatively fast or their best days for this game are going to now be behind them. I've been unsuccessful at bringing a single friend over from GW2 or WoW strictly due to the poor performance of this game with respect to latency and response times. As an aside, they also often site the 2.5s GCD in this game compared to the significantly faster GCDs of games like GW2, Rift, and WoW (and most MMOs for that matter). And while I personally don't mind the 2.5s GCD and I feel that folks can easily get used to it, when you combine this slower pace of play with poor server conditions, you have a recipe for a very, VERY negative perception about how this game plays. -
09-01-2015, 02:41 PMRaistIt doesn't apply to everyone you've spoken to all the time. Guess you missed that I have no such issues...and there are many others like me. You apparently have issues with how your data is traversing the internet. There are MANY others here who also don't. The reason is simply that they are routed differently. Some get it from their ISP's efforts while others are doing it via a VPN.Quote:
Originally Posted by Synovius
And no, I don't rely on a VPN for this game. I used one off and on during the first months of the 2.0 relaunch, mostly as a tool for demonstration to my ISP's tier2 and tier3 techs when things would go south. Eventually they stopped going through the checklist every time I sent them a dump of traces, pings, and modem logs and just went to work tracking down the problem and fixing it for me.
Ultimately we narrowed down some localized issues to failing hardware at the head end--was actually a device that had been in play since Clinton was in office and was on it's last leg. Still haven't been put back on a dedicated gateway for Florence yet...but the Conway one isn't too bad. Most days.
They still have some upstream issues to address in the Carolinas...but we have seen much improvement in the last year. But we have to stay on them though...it is Time Warner after all. Their forums are constantly lit up with complaints...a large part of it is upstream issues(though there are still a lot of local channel alignment issues---still a lot of outdated hardware in play). The biggest change though is likely bound to us pushing them to keep us away from Level3 as much as possible. Seems to be one of the more common weak links in the routing.
But...if people aren't willing to look at the facts or listen to reason....guess we may not see any lasting dramatic improvements for them anytime soon.
EDIT:
Out of curiosity...I booted the laptop to take some snapshots. Here is some quick data on the route I take to my lobby server for Midgard (trying to avoid spamming the game servers themselves for these thing), followed by the routing to the two WoW servers in North America for a comparison. Note the GeoData for these IP's are Montreal, New Jersey, and California from a laptop on Wi-Fi in South Carolina:
Code:C:\Windows\System32>tracert neolobby02.ffxiv.com
Tracing route to neolobby02.ffxiv.com [199.91.189.74]
over a maximum of 30 hops:
1 1 ms 2 ms 1 ms LPTSRV [10.10.100.1]
2 26 ms 30 ms 27 ms cpe-75-176-160-1.sc.res.rr.com [75.176.160.1]
3 26 ms 23 ms 20 ms cpe-024-031-198-005.sc.res.rr.com [24.31.198.5]
4 13 ms 11 ms 15 ms clmasoutheastmyr-rtr2.sc.rr.com [24.31.196.210]
5 23 ms 29 ms 27 ms be33.drhmncev01r.southeast.rr.com [24.93.64.180]
6 28 ms 29 ms 30 ms ge-0-3-0.chrlncpop-rtr1.southeast.rr.com [24.93.64.173]
7 31 ms 33 ms 34 ms bu-ether34.atlngamq46w-bcr00.tbone.rr.com [107.14.19.48]
8 31 ms 30 ms 29 ms 0.ae1.pr0.atl20.tbone.rr.com [66.109.6.177]
9 30 ms 28 ms 31 ms te0-0-0-10.ccr21.atl02.atlas.cogentco.com [154.54.12.109]
10 31 ms 30 ms 30 ms be2050.ccr41.atl01.atlas.cogentco.com [154.54.0.165]
11 43 ms 43 ms 44 ms be2112.ccr41.dca01.atlas.cogentco.com [154.54.7.157]
12 47 ms 47 ms 50 ms be2148.ccr41.jfk02.atlas.cogentco.com [154.54.31.118]
13 52 ms 53 ms 51 ms be2106.ccr21.alb02.atlas.cogentco.com [154.54.3.50]
14 56 ms 57 ms 56 ms be2088.ccr21.ymq02.atlas.cogentco.com [154.54.43.17]
15 61 ms 60 ms 61 ms 38.122.42.34
16 58 ms 57 ms 58 ms 192.34.76.10
17 63 ms 65 ms 58 ms 199.91.189.242
18 59 ms 58 ms 59 ms 199.91.189.74
Trace complete.
Ping statistics for 199.91.189.74:
Packets: Sent = 10, Received = 10, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 54ms, Maximum = 58ms, Average = 56ms
Code:C:\Windows\System32>tracert 12.129.222.10
Tracing route to 12.129.222.10 over a maximum of 30 hops
1 1 ms 1 ms 1 ms LPTSRV [10.10.100.1]
2 33 ms 29 ms 29 ms cpe-75-176-160-1.sc.res.rr.com [75.176.160.1]
3 25 ms 27 ms 31 ms cpe-024-031-198-005.sc.res.rr.com [24.31.198.5]
4 22 ms 23 ms 35 ms clmasoutheastmyr-rtr2.sc.rr.com [24.31.196.210]
5 27 ms 25 ms 28 ms be33.drhmncev01r.southeast.rr.com [24.93.64.180]
6 29 ms 28 ms 29 ms ae14.chrlncpop-rtr1.southeast.rr.com [24.93.64.25]
7 32 ms 33 ms 33 ms bu-ether24.atlngamq46w-bcr00.tbone.rr.com [107.14.19.18]
8 38 ms 37 ms 37 ms 66.109.1.255
9 32 ms 31 ms 39 ms 0.ae2.pr1.atl20.tbone.rr.com [107.14.17.188]
10 29 ms 31 ms 30 ms ix-14-0.tcore1.A56-Atlanta.as6453.net [64.86.113.37]
11 33 ms 33 ms 33 ms 192.205.36.217
12 47 ms 49 ms 49 ms 12.122.29.54
13 47 ms 49 ms 48 ms cr2.wswdc.ip.att.net [12.122.1.174]
14 47 ms 47 ms 49 ms 12.123.250.117
15 * * * Request timed out.
16 * * * Request timed out.
17 * * * Request timed out.
18 ^C
<I stopped the trace, as I hit their firewall as detailed in their guide. Also unable to ping it>
Notice a few very important things about this data:Code:C:\Windows\System32>tracert 199.108.62.157
Tracing route to 199.108.62.157 over a maximum of 30 hops
1 1 ms 1 ms 1 ms LPTSRV [10.10.100.1]
2 29 ms 24 ms 28 ms cpe-75-176-160-1.sc.res.rr.com [75.176.160.1]
3 33 ms 27 ms 29 ms cpe-024-031-198-005.sc.res.rr.com [24.31.198.5]
4 12 ms 13 ms 14 ms clmasoutheastmyr-rtr2.sc.rr.com [24.31.196.210]
5 23 ms 24 ms 24 ms be33.drhmncev01r.southeast.rr.com [24.93.64.180]
6 28 ms 29 ms 30 ms ge-0-3-0.chrlncpop-rtr1.southeast.rr.com [24.93.64.173]
7 31 ms 32 ms 33 ms bu-ether34.atlngamq46w-bcr00.tbone.rr.com [107.14.19.48]
8 37 ms 29 ms 38 ms 66.109.1.255
9 31 ms 32 ms 32 ms 0.ae3.pr1.atl20.tbone.rr.com [107.14.17.190]
10 30 ms 28 ms 33 ms ix-14-0.tcore1.A56-Atlanta.as6453.net [64.86.113.37]
11 48 ms 45 ms 45 ms if-13-2.tcore1.AEQ-Ashburn.as6453.net [64.86.113.10]
12 43 ms 44 ms 46 ms if-2-2.tcore2.AEQ-Ashburn.as6453.net [216.6.87.1]
13 47 ms 48 ms 47 ms 192.205.34.245
14 47 ms 46 ms 49 ms cr1.wswdc.ip.att.net [12.122.113.42]
15 46 ms 44 ms 45 ms 12.123.250.37
16 * * * Request timed out.
17 * * * Request timed out.
18 * * * Request timed out.
19 ^C
<again, hitting their security...so can't ping this one either>
1.) No more XO.net and no Level3 hops at all in any of those routes...these were frequently fingered as major offenders in our testing.
2.) Fairly tight, and consistently low response times throughout the entire route.
3.) No circular/redundancies in the routing (we frequently see traces return the SAME level3 hops back to back in a single trace)
4.) Response times are mostly in keeping with distance--no major spikes disproportionate to the distance
(note that I go to Conway/Myrtle Beach and then to Columbia first--problems in Florence has resulted in strange band-aids. At one point, I went to Hilton Head first. Believe it or not...in both cases this sling-shotting proved much more stable)
Basically...these are "clean" and consistent routes. As a result, my gameplay is also stable and responsive...even when people are streaming to the Roku's.
Oh yeah... and barring ONE incident when TWC thought they caught an exception between Ormuco and SE---we haven't been in contact with them at all.
This was managed through Time Warner.
Edit 2:
0.o just after posting that data, they rebooted my modem....and my route has now changed. Must be some maintenance going on or something---I'm peering with Cogent now instead of TATA:
Cogent isn't typically as solid as it has been with TATA (will often get some 150-200ms hiccups once in a while with Cogent)---but it is NOTHING like the massive spikes we see with Level3 in play...talking full on 1000+ starred events. As in full-on one second pauses (or longer) with Level3 when they misbehave (and that's not tracerts mind you--straight pings).Code:Tracing route to neolobby02.ffxiv.com [199.91.189.74]
over a maximum of 30 hops:
1 1 ms 1 ms 1 ms LPTSRV [10.10.100.1]
2 33 ms 29 ms 27 ms cpe-75-176-160-1.sc.res.rr.com [75.176.160.1]
3 24 ms 32 ms 30 ms cpe-024-031-198-005.sc.res.rr.com [24.31.198.5]
4 12 ms 12 ms 25 ms clmasoutheastmyr-rtr2.sc.rr.com [24.31.196.210]
5 26 ms 22 ms 27 ms be33.drhmncev01r.southeast.rr.com [24.93.64.180]
6 26 ms 29 ms 30 ms ge-0-3-0.chrlncpop-rtr1.southeast.rr.com [24.93.64.173]
7 31 ms 32 ms 31 ms bu-ether34.atlngamq46w-bcr00.tbone.rr.com [107.14.19.48]
8 33 ms 52 ms 73 ms 0.ae1.pr0.atl20.tbone.rr.com [66.109.6.177]
9 31 ms 29 ms 30 ms te0-0-0-10.ccr21.atl02.atlas.cogentco.com [154.54.12.109]
10 30 ms 30 ms 31 ms be2050.ccr41.atl01.atlas.cogentco.com [154.54.0.165]
11 43 ms 44 ms 43 ms be2112.ccr41.dca01.atlas.cogentco.com [154.54.7.157]
12 49 ms 47 ms 48 ms be2148.ccr41.jfk02.atlas.cogentco.com [154.54.31.118]
13 52 ms 52 ms 52 ms be2106.ccr21.alb02.atlas.cogentco.com [154.54.3.50]
14 56 ms 57 ms 58 ms be2088.ccr21.ymq02.atlas.cogentco.com [154.54.43.17]
15 70 ms 72 ms 72 ms 38.122.42.34
16 55 ms 59 ms 57 ms 192.34.76.10
17 56 ms 57 ms 59 ms 199.91.189.242
18 56 ms 57 ms 57 ms 199.91.189.74
Trace complete.
Again... Time Warner just did this. Not Square-Enix. -
09-02-2015, 04:44 AMSynovius
Out of curiosity, Raist, have you played WoW at all? I think if you had it would be immediately apparent how much more responsive and crisp the gameplay is. And, again, I don't even like the game now because I found WoD to be boring. However, there is absolutely no denying how responsive the game is compared to FFXIV - and this is even while playing on a server located across North America from me.
Either way, I will do some pinging and tracert when I get home here shortly to Gilgamesh and report the results here. -
09-02-2015, 05:24 AMRaist
Been at this for nearly 20 years now...back to the days of dial-up and the introduction of residential broadband to our area that only offered 128k to 2mbit bandwidth plans that costed a premium of up to $70 a month---and that was bundled or yearly contract pricing. Online space flight sim, FPS, RTS, MMORPG, video conferencing/VOIP, regular streaming A/V....been there and done it all. Hosted content off old x86 platforms that would be a toaster compared to today's standards (were a lot of games where you hosted missions and maps locally on your machine--the game server was just for matchmaking).
We have always had to wrangle with latency across the internet. Still the same now as it was then...often the great equalizer (or punisher depending on how good/bad yours is). Content can be served on the best and most expertly optimized hardware at the endpoint...and a flaky router in Cleveland can kill a group of users' service in a heartbeat.
The servers and this game are plenty responsive enough. We have frequently found that unreliable sessions across the internet cause a major chunk of the issues. This has been repeatedly affirmed on these forums...in general, people with good transit have a good gaming experience. Granted there have been issues on SE's side in specific cases, but in the general scope of things those are the exceptions. The general consensus is bad session means bad game play. -
09-02-2015, 05:51 AMSynovius
That still doesn't address the idea that WoW's gameplay is extremely crisp and responsive even when going across a continent while FFXIV's responsiveness is very, very poor. Just a few examples from my latest DF run in Fractal Continuum:
- On the jump pads in the last room, without sprint popped, I am able to run completely over the glowing arrow by 3-4 steps before I actually get launched to the next platform.
- I had a Firestarter proc that I saved until there was a full 1s left on the buff. When I went to use it, it "consumed" the Firestarter proc and the Fire III animation went off but no damage was done and I did not receive my 3 AF stacks for it.
- Twice during the second boss (the minotaur), I was able to complete a Fire IV cast while only a bit beyond halfway through it despite the cast of "Deafening Groan" going off which knocks everyone back and should have interrupted casting.
The above are things I went ahead and allowed to happen for the purposes of demonstration and ALL are things I have to personally plan ahead for when playing this game but NEVER would have to plan ahead for while playing WoW. In raids, I know that I need to use Firestarter with 2s or more left on it or I won't get credit for it. I know that, if there's any kind of reactive AoE or "collision" that needs to take place, I should run to the correct spot and stand there for a solid 1-2s to ensure the server registers that I was there. In raids, I use the poor performance to my advantage to complete casts that realistically should have been interrupted or not gone off due to a phase transition.
What's more is that I'm not the only one doing this. Our entire raid does this every raid night. Do you know how many times we wiped to AS3 simply because the server didn't pick up that our off-tank and healer were standing on top of each other for a full second yet the damage-increased debuff didn't trade from the off-tank to the healer? We went as far as having a waymark as a designated spot and I watched for pull after pull after pull where the appropriate healer and off-tank would run to the A marker and literally stand on top of each other for a full second or two only to sometimes see the buff not be transfered and thus cause an immediate wipe.
This is not a client-side issue that I or anyone individually can resolve. At worst, this is a server-side issue that Square Enix needs to address or, at best, this is an ISP issue that Square Enix needs to address directly with the ISPs.
Like you, I've been at this for over a decade. I'm 33 and have worked as a web developer since 20 years of age. I love this game but if the poor performance of the servers doesn't get addressed, the best days for this game are behind it.
FYI: I just ran a tracert just like yours and my route is even faster than yours...
Code:Tracing route to 199.108.62.157 over a maximum of 30 hops
1 1 ms <1 ms <1 ms CISCO52642 [192.168.1.1]
2 <1 ms <1 ms <1 ms home.Home [192.168.200.1]
3 17 ms 17 ms 21 ms RO1-DSL-208-102-248-1.fuse.net [208.102.248.1]
4 17 ms 17 ms 17 ms 172.17.74.18
5 18 ms 17 ms 17 ms EV-ZT-1.EVE2.core.fuse.net [216.68.14.58]
6 17 ms 17 ms 17 ms EVE2.EVE1.core.fuse.net [216.68.14.44]
7 18 ms 17 ms 17 ms te0-0-2-2.nr11.b016343-1.cvg02.atlas.cogentco.co
m [38.122.234.101]
8 18 ms 18 ms 18 ms te0-0-1-1.rcr11.cvg02.atlas.cogentco.com [154.24
.0.125]
9 20 ms 20 ms 20 ms te0-2-0-0.rcr21.ind01.atlas.cogentco.com [154.54
.84.166]
10 23 ms 24 ms 23 ms te4-3.ccr01.sbn01.atlas.cogentco.com [154.54.27.
82]
11 28 ms 27 ms 27 ms te0-10-1-0.ccr41.ord01.atlas.cogentco.com [154.5
4.80.149]
12 27 ms 27 ms 27 ms be2216.ccr41.ord03.atlas.cogentco.com [154.54.24
.202]
13 27 ms 31 ms 27 ms att.ord03.atlas.cogentco.com [154.54.12.86]
14 30 ms 27 ms 27 ms cr1.cgcil.ip.att.net [12.122.133.34]
15 33 ms 32 ms 32 ms 12.123.251.17
16 * * * Request timed out.
17 * * * Request timed out.
18 * * * Request timed out.
19 * * * Request timed out.
20 ^C
-
09-02-2015, 06:11 AMRaist
You are sort of answering your own question (notice you are also not using Level3 there? Don't see xo.net either...).
WoW is by no means immune to the exact same phenomenon you are describing for this game, so please stop pretending like it is some perfect contrast to this game. The fact that you appear to have solid sessions to their servers and also have solid game play with them is more or less proving the point--you have good routing to them, and that is reflected in the response times. The internet is full of complaints over bad latency knackering up WoW just like it does for XIV, XI, GW2, LoL,....even Netflix and YouTube for heavens sake. There is a reason Blizzard has that post detailing how to test for latency after all...many other services will have similar articles/tools posted as well for conducting tests. It is because it is a KNOWN problem that must be investigated and ruled out of the equation.
There are known address groupings that habitually demonstrate stability issues associated to the very symptoms you are describing. Level3 is not the only ISP that has been flagged for it in the past, but it shows up quite often in forums across multiple services---including WoW. And specifically those car2 lines in the Montreal area are sort of notorious for it.
Go ahead and google latency issues for WoW if you don't believe me. I'll even provide a link for a very simple search for you:
https://www.google.com/?gws_rd=ssl#q=latency+issues+WoW
No matter what is going on at either the client or server side--if you have a flaky route, a client-server based service is going to be negatively impacted. PERIOD. If there are signs of issues in routing, they need to be addressed, otherwise you are going to be open to the negative affects that can be triggered by those problems. This has been demonstrated often in these forums as well as elsewhere in the past. Try hosting a space combat skirmish/dogfight with north of 130ms latency for everyone else, than do it again when the latency is 60-80 for everyone. The players will see and "feel" a marked difference in accuracy and such. The only thing that changed is the latency.
Oh, and by the way... Time Warner has switched me back to peering with TATA again for getting to Ormuco. Just saying....routing policies can be amended to address a lot of routing issues once they are properly investigated. -
09-02-2015, 06:34 AMRaist
Glancing back through the thread just now, noticed something I missed earlier:
Begs the question if you are missing the cast bar factor on these moves. The animations and ground markers are not your indicators of when the move is firing...it is not notify you of the action pending, and where it will land. Your timing needs to be aligned to the progression on the casting bars. That is the indicator of when the server is calculating things. It was the same in XI...it is how we have always pulled off stutter stepping and such.Quote:
Just an hour spent on ANY WoW server - even one across the United States from you (I live in EST and often played on PST servers) - yield crispness and seemingly immediate response times where moving out of an AoE a the very last split second is perfectly fine, moving into an AoE buff buffs you literally the split second you arrive in it and a Warrior's Charge ability tracks you every point of the way from your source to destination - so much so that you can be stunned mid-charge at any point and you will freeze immediately at that location.
-
09-02-2015, 06:37 AMSynovius
I appreciate your willingness to help but this is 100% not an issue on my end nor is it something I can address. It was this way when I played on Adamantoise, on Behemoth, and now on Gilgamesh. My entire raid suffers from this nightly. With over a decade of MMO gaming, normal "lag" is very noticeable for me. The difference between 200ms and 50ms on, say, Mal'ganis (US) in WoW is extremely noticeable. That 150ms is the difference in your Cheap Shot (a Rogue stun) going off seemingly instant or for their being a slight delay upon pressing the hotkey and seeing the animation and resulting stun take place.
What we're all experiencing in FFXIV is not "lag" in this sense. Without looking at the game's networking and graphics code I couldn't begin to tell you what the culprit is. It could be WoW uses substantially better lag correction algorithms to address this very concern. Again, if I play on a WoW server and I step into an AoE buff, within 50ms (that is, to say, basically instantaneous to human perception) that buff is on me. When I step into an AoE buff in FFXIV, a solid second often passes before that buff takes effect. This has trickle effects on gameplay in FFXIV everywhere.
The best I can explain it (and the last of my explaining as I'm beating a dead horse here and you clearly just want to sweep this under the rug as "eh it's a non-issue") is that playing FFXIV at 100ms feels like playing Unreal Tournament 2k4 at 400ms while playing WoW at 100ms feels like playing Unreal Tournament 2k4 at 100ms. One is very precise and very responsive even given the exact same latency measured in ms while the other is not.
We can call this a "routing issue" and we can sweep this under the rug as a "non-issue" but it is and it needs to be addressed or this game has peaked and will begin its steady decline.
I'm very aware of the fact the particle effects and animations in this game do not line up properly in many cases with the actual effect. Take Titan's Landslide, for example. The ground colors disappear but the effect goes out after they are gone. This is fine although I hope you'll recall the massive !@#%storm that was hundreds of players making videos and complaining about being knocked off the platform in Titan HM and EX despite having long-since left the Landslide effect.Quote:
Begs the question if you are missing the cast bar factor on these moves. The animations and ground markers are not your indicators of when the move is firing...it is not notify you of the action pending, and where it will land. Your timing needs to be aligned to the progression on the casting bars. That is the indicator of when the server is calculating things. It was the same in XI...it is how we have always pulled off stutter stepping and such.
-
09-02-2015, 07:04 AMRaist
Again....need to pay attention to the progress of the casting bar.
And not trying to "sweep it under the rug". Been working on the latency issue specifically for this game for well over a year now. Not just for myself...but more so in trying to get people on track to get their issues addressed. My issues pretty much got a permanent fix of sorts back at the first of the year, but occasionally we have to revisit things because the traffic patterns sometimes shift faster than TWC catches up and I have to shoot them an email. But, I haven't had to do that since sometime back around Easter....they've been pretty spot-on for the most part with catching it. A few times I've started generating the reports to email them, and then the modem reboots on it's own and everything is cleared up again before I actually get around to typing up the email.
A lot of people complaining about the very same things you have been were suffering from lag (that includes the landslide dilemma)...but it is beginning to sound like your issue may actually be two-fold, or perhaps just that one mechanic that you may be overlooking---the proper indicator of when an action is taking place. The proper indicator is the casting bar. If your timing is aligning properly there, then it may simply be attributable to your latency.
People with good latency that are moving appropriately according to the proper indicators seem to be doing fine.
Here is a little video I found on YouTube that demonstrates it somewhat. Pay attention to the progress bar compared to what is going on on the screen. The same principles apply to the markers and such you are complaining about:
https://www.youtube.com/watch?v=aAGNizT-6xo -
09-02-2015, 07:48 AMSynoviusI really do appreciate your willingness to help but, at the risk of sounding conceited, I've been raiding for a very long time and am currently in world top raiding FC and my static in particular is working on AS4 atm. We are all very, very aware of castbars, when to move, when not to move, how to take advantage of the horrible performance, etc. This is not a player issue here. I'm fully aware that when I get down to 1s or even 1.5s on my Firestarter, I shouldn't use it because it's going to throw me into the GCD without any AF stacks or even the damage from the Fire III despite the animation and particle effect going off. You are right that, in FFXIV (and not in WoW for example), we have to react perhaps a full second or two before a cast finishes because, if we don't, the game will consider our characters as being present at our previous location when the cast goes off instead of our new location we are standing at when we get hit by whatever ability came out. Lots of abilities in this game suffer from the same issue. For example, to bring back the Charge ability for WoW Warriors, it is a long-distance charge similar to Shoulder Tackle by Monks in FFXIV. In WoW, however, if you are really good, you can time a stun mid-charge even though the entirety of the charge's animation takes place over about 1-1.5 seconds. If you were to stun this Warrior with a very well-timed stun mid-charge, the Warrior would be stunned in place IMMEDIATELY where you stunned them. Contrast that with "charge" type mechanics in FFXIV. Taking our own job of BLM for example, if you are standing in an AoE and you use Aetherial Manipulation to someone who is outside of that AoE, even if you are visually WAAAAAAAAY out of the AoE (by 200-300ms at least), the game will still consider you as being present at your source location instead of where you actually, visually were when the AoE went off. I've died like this in the past many times wherein I've Aetherial Manipulated to someone and after my movement has already stopped and I'm standing next to my AM target, I die and my corpse is lying some 25 yalms away from the AoE that exploded. It does not matter if I have flawless routing directly to Gilgamesh for this with 10ms latency. The same effect would occur because the game does not seem to update positions as often as other MMOs. The same issue with AM occurs with Dragoon jumps, Monk Shoulder Tackle, Ninja teleport, DK charge, and any positional abilities like these.Quote:
Originally Posted by Raist
And, FWIW, the video you showed about "casting while moving" is something I do every single raid. I am always "stutter stepping" between my casts when I have to move. The fact that this is possible does not mean it's a good thing. If it were the case that this were possible solely because code existed to allow for movement yet complete casts after a certain % of the cast has completed, that would be one thing. But the reason this is possible is not solely due to that - it's due to the fact that even if I move when my Fire IV is at only 70% full, the cast will usually complete successfully.
Further, how is it that you think I am suffering from "lag" in this game and that's actually my issue when you say you don't experience this performance issue at all yet my connection to the server is better than yours both in total hops as well as latency? My impression that you are simply conditioned to accept and deal with the poor performance whereas many of us who came to FFXIV from other MMOs were immediately (and still) blown away by the poor performance of this game's servers. -
09-02-2015, 10:36 AMRaist
That trace was to a WoW server... not an FFXIV one.... not seeing where your route for XIV is performing any better. In fact, haven't seen a trace from you to an FFXIV server at all.
The game automatically polls multiple times per second...every 300ms in overworld, forget what it was in instances but do recall something about they were going to increase the frequency during instances a while back. It also sends positional updates when you submit actions as well.
The reason stutter stepping works is we are sneaking between the automatic updates. Once you submit an action, the server is sort of on auto-pilot for that action until something happens server side to interrupt it--be that something calculated server-side, or something that happens in our client (positional update shows we moved, submit an action from another location that hits the server before the casting finishes on the server side, etc.). If you get in synch with the delays, you can pull off some crazy stuff. Not only does it allow us to cheat on our movement...it also affords us a means to force awareness of our movement as well.
The reason why it might be a latency issue is that if you feel you are in synch with the delays but are still getting hit, then your latency may be worse than you thought. If you consistently clear the target area 50% through progress just fine, but do not when it is 75% through... then you have unwanted latency. In a more perfect world, you would likely be able to stay in the target area on up to 90 or even 95% (have to allow for a slight bit of input and processing lag still). The trick is figuring out just where that threshold is and adjusting to it if the window is really tight...which is what a lot of people had been failing to do. Many weren't even paying attention to the cast bar...they were talking just like you were earlier and judging it by the animations or the state of the marker, either one of which can be out of synch by varying degrees under different conditions---the progress bar is a much more consistent item to synchronize with.
And as I have stated multiple times...I don't experience these performance issues like you do. Yes, there can be a noticeable delay from hitting a button and seeing the feedback on the screen---but that can happen with any service that is relying on a response from a server. Heck, you can even feel it sometimes in a standalone environment that doesn't involve sling-shotting to a server and back. An eye blink takes right under 1/3 a second on average. 125ms is 1/8 a second...so it would require extremely low latency for a more true instant response feel from a server nearly 900+ miles away. What little delay that is perceptible to me simply is not game-breaking for me. I can dodge AOE's and keep Eno up just fine in all kinds of crazy situations--even in the expert roulettes. When I do start to actually "feel" any real lag, it often is hardly even annoying really until I start crossing about the 200ms mark a lot if any issues start creeping up (by the time I start crossing that line though, my ISP is usually on it and things get changed---sometimes my modem reboots mid-instance even, but it happens so quick I manage to get back in quickly and finish).
Basically, again as was stated earlier, typically when the route is good... the game tends to be good. This has been shown to be true many times for others as well. So, all things being "equal" at the other end (same network, same hardware, same software, etc. on their end)---if a lot of people are having it good, and you are not, what is the difference? In a LOT of cases we have seen issues present in the routes for the troubled players that simply do not exist for those that are not having a hard time.
At that point, it seems pretty elementary my dear Watson. Focus on the signs of smoke to see if there is fire, and if found snuff it out...get the route stabilized and see if game play improves. For many, the quick and dirty fix was testing a VPN...when that resolves it, it pretty much clenches it that something within the networks between the user and the server is causing issues. Knowing that, you have a starting point for investigation....once one can rule out their local hardware, that should involve one's ISP as they are the ones that first take over on the other side or your private network.
Many have followed such logic and pushed for the escalation, gotten it, and have been met with success. Regardless if it makes sense or not, it sets a precedent of sorts. Bad routing negatively impacts gameplay. As long as that condition persists, gameplay will be negatively impacted. Thus...it needs to be investigated and flushed out if suspected.
Oh yeah... also.... Fuse.net, Cincinnati. Ohio has been shown in the past to run into issues with congestion and various other issues in the past---circular routing, bad channel alignments and signal ingress, screwy weather patterns---you name it, they have likely experienced it if it can negatively impact a network. And it tends to get progressively worse the closer you get to the lake. Cleveland has had it's share of nightmares in recent years. Just saying, there is a lot of history for your region...and not all of it favorable for getting to the other side of the lakes.
Looked them up...appears to be Cincinnati Bell under the hood:
Cincinnati Bell downdetector page
(among other things, has links to their social media, main website, etc. in case support is needed) -
09-02-2015, 11:49 AMTrain88
If your router is older than 2 years, this may be signs it is dying.
Since you have a pretty reliable way to generate this lag, you may want to connect the modem to the computer if at all possible. If you do not get lag to have solved your problem. If you still get lag you may want to phone your isp and ask them to verify your signal quality. This signal quality is their responsibility.
Also if you're playing over wifi, don't play over wifi. The nextdoor granny's microwave oven will kill your signal. If there is anything you can do to improve your wifi signal it may help a lot, assuming there is no possible way for you to use a cabled connection.
That's all i got for internet connection, hope it helps. -
09-02-2015, 11:54 AMSynovius
I'm not sure what to say, honestly. I posted a tracert to Gilgamesh that showed the connection at 39ms. I'm fairly certain my latency is actually very low - sub 100 low. I'm not sure what trying to get a better route would accomplish as I already have very low latency.
Maybe I'll make a video and, in it show both WoW and FFXIV and my latency to the server in both and then I'll demonstrate exactly what I mean about how crisp things seem in WoW (even with a much higher latency) and how sluggish and unresponsive things seem in FFXIV. -
09-02-2015, 12:00 PMRaistThat is a WoW address, not FFXIV.Quote:
Tracing route to 199.108.62.157 over a maximum of 30 hops
Edit:
Was just farting around in some looking glasses and spotted some packet losses from various paths that went through the Toronto area. Did some poking around some more and there's some "funkiness" in general around Lake Erie and Lake Ontario from all kinds of different vectors. Losses weren't earth shattering by any means but was getting a lot of spiking in the 350-650 ranges whenever cutting through some of the Toronto corridors. A good bit of traffic from the western side of New York can sometimes get shunted that way if things aren't fairing well on the other side of Syracuse or Albany.
So there could be some weird delays happening along the way depending on how you get routed. Perhaps just enough for TCP's error correction to trigger a retrans here and there. Could be a tricky one to catch though....took almost 2 minutes for it to show up when I let a ping from the laptop to neo2 to just run continuously. Could be nothing... could be the start of something. Time will tell. -
09-02-2015, 12:13 PMSynovius
Oh my bad. Do you know the IP then for Gilgamesh or a server from the same datacenter?
BTW, our SCH in the raid lives within a 1hr drive of the data center for Gilgamesh and he experiences the exact same issues with responsiveness that I do.
EDIT: I pinged the NA data center and it my latency is 32ms.
http://i.imgur.com/ArutVZh.png
I then did a tracert to the datacenter and it came back at 37ms. -
09-02-2015, 01:06 PMRaist
184.107.107.176 isn't an XIV server either...that is an EIDOS webserver. It hosts sites mostly for IP's from that leg of Square-Enix. You get the same address if you ping things like eidos.com, justcause2.com, etc.---something like a dozen sites point there. That server is actually hosted on i-Web (not Ormuco) and in a different section of Montreal. If I remember right, there is actually a waterway between them.
The world services are actually clusters, and we get directed to a front end that has a bank of IP's to use. So, not everyone on the same world will necessarily be on the same IP address. I've tracked up to 3 different addresses in use at the same time amongst posters here who are also on Midgard. The assignment can change periodically (I've gone from 25 to 31 to 30 back to 31...have to check it periodically). The only addresses that have remained consistent have been the lobby servers, which are registered with DNS names (neolobby01.ffxiv.com through neolobby06.ffxiv.com, currently the odd ones are Japan and even are Canada, and Aether has consistently been neolobby02 on .74 for me).
Best way to know for certain is to look in the network sections of Resource Monitor (resmon.exe from search/run box) after you are fully logged into the game.
As for your friend that lives close to Montreal...proximity doesn't always equate to excellent connections when it comes to the internet. You can get routed to another city and then come back just to get to a server in your same town---or even the same street for that matter. Case in point:
That is a trace to a box in our county services department...one I physically delivered during the county's implementation of online tax payment services years ago. I can drive there in 8 minutes...but it is bouncing between South and North Carolina before it comes back to Florence, SC. Geolocation data is all messed up for it (Time Warner really needs to get it's act together)--some records state Myrtle Beach (business class service comes out of there sometimes, others will show up as Columbia), so the routing gets all jumbled up.Code:Tracing route to XXXXXXXXXXX <oops...they might not like someone posting the actual address>
over a maximum of 30 hops:
1 1 ms <1 ms <1 ms LPTSRV [10.10.100.1]
2 46 ms 21 ms 41 ms cpe-75-176-160-1.sc.res.rr.com [75.176.160.1]
3 22 ms 29 ms 30 ms cpe-024-031-198-005.sc.res.rr.com [24.31.198.5]
4 14 ms 15 ms 13 ms clmasoutheastmyr-rtr2.sc.rr.com [24.31.196.210]
5 28 ms 24 ms 25 ms be33.drhmncev01r.southeast.rr.com [24.93.64.180]
6 23 ms 42 ms 24 ms 24.93.67.99
7 26 ms 28 ms 26 ms ten3-0-0.fyvlncr-pe-rtr01.southeast.rr.com [24.93.73.42]
8 27 ms 27 ms 29 ms ten3-0-0.fyvlncr-p-rtr01.southeast.rr.com [24.93.73.81]
9 28 ms 30 ms 28 ms ten4-0-0.chrlncsa-pe-rtr01.southeast.rr.com [24.93.73.65]
10 26 ms 27 ms 26 ms ten2-0-0.clmascmhe-p-rtr1.southeast.rr.com [24.93.73.53]
11 27 ms 28 ms 31 ms lag3.twcc.clmascmhe-a2701.sc.rr.com [24.27.255.251]
12 * * * Request timed out.
13 ^C
<hit their firewall...it is only open on specific ports>
We've seen some issues with both alter.net (part of Verizon) and tinet/SPA up there in the past, so if he/she is coming in on one of their lines it could be flaky. Come to think of it...idk if the FiOS users up there ever got any decent relief either. They were lighting up forums last month big time for various things across the north east and a few other regions. Remember specifically the north east though, as the level3 exchanges up there were also in play, which have proven to be a big source of frustration in general for a long time now.
Toronto in general started showing weirdness tonight too. Wasn't any particular ISP though---the big three were all acting up (TATA, Cogent, Level3). -
09-02-2015, 02:18 PMSynovius
While I'm a web developer and not a networking expert, I do understand everything you're discussing (although you no doubt know a great deal more about it than I do or ever will).
What you're saying, however, is still not adding up to the fact that literally everyone I have ever played this game with and everyone I currently play this game with, complains about the same thing: the lack of responsiveness from the game servers evidenced by raid environments where every player has to be exceedingly deliberate with every action and plan to play "around" the latency issue. In many respects, the strategies we implement in a raid environment are based around latency. Taking AS3 for example, it should be enough to say that, when the off-tank gets the damage up debuff from Digititis, he and a healer should simply run into each other and thus transfer the debuff. What we've had to do, however, is mark a designated spot and both of them, literally, run to that spot and stand within each other's hitbox for a solid 1-2s until the debuff finally transfers from the off-tank to the healer. Even if this was strictly a latency issue from the two players' PCs to the data center, it still would not account for the majority of this poor responsiveness.
Again, I repeat, I have raided in WoW at 200, 300, even 400ms before. The latency to the game server is listed right in the base UI for you. Even at such terrible latency, the responsiveness and crispness of the gameplay therein was a substantial improvement over what I see on a day to day basis in FFXIV.
If there is something specific and actionable you think I should try, I'm all for it but I have a feeling that, were it this simple, there wouldn't be such a vast collection of players suffering from the exact same issue and it wouldn't be so consistent as to be a 100% certainty every time we all log into the game. -
09-02-2015, 10:52 PMJudge_Xero
I don't have much other MMO experience but IMO the responsiveness of FFXIV is pretty lacking.
I play at 30ms just 600 km from the data center. I still have to use a VPN because anything over 10ms makes the input delay almost unbearable for me.
I think the example of having to pass debuffs is great because you are relying on the connection of two players on top of the update speed of the servers which really amplifies where the servers are lacking (update intervals)
I will be really interested how things will be when the EU servers go live. Hopefully SE can crank things up a bit when the subs are spread out over an additional server.