Lack of SE Response to Lag

[Duuude007]

Obviously the moose threatened SE and they had no choice but to put them there. CLEAN HANDS.

Seeing that hundreds of thousands more people can play without lag caused by playing from japan datacenter, its a failure, yes?

[Allyra]

Yep - If it was purely SE, wouldn't XI's players also be suffering the same "unplayable lag"?

No one has said it's 100% SE's fault, stop exaggerating for one. Two, you can't compare the two because:

They are not using the same hardware
The netcode was not made by the same exact people
FFXI was specifically built for the ps2
FFXI is not as data intensive
FFXI doesn't even send the same kind of data
FFXI is 10 years old which means lag could have been fixed overtime
Rubberbanding does exist in FFXI (If you played it, I guarantee you would have seen several times people's toons running sporadically to catch up to you)
FFXI doesn't have "unplayable lag" because the game isn't dodge intensive, and you do not get locked out of content over lag.

If there is faulty "insert X" (netcode, hardware, data center placement, etc.) in one, it's not going to guarantee there is faulty "insert X" in the other.

Seeing that hundreds of thousands more people can play without lag caused by playing from japan datacenter, its a failure, yes?

I have no idea what you are asking, but the JP are having problems,too.

[Marishi-Ten]

No one has said it's 100% SE's fault, stop exaggerating for one. Two, you can't compare the two because:

They are not using the same hardware
The netcode was not made by the same exact people
FFXI was specifically built for the ps2
FFXI is not as data intensive
FFXI doesn't even send the same kind of data
FFXI is 10 years old which means lag could have been fixed overtime
Rubberbanding does exist in FFXI (If you played it, I guarantee you would have seen several times people's toons running sporadically to catch up to you)
FFXI doesn't have "unplayable lag" because the game isn't dodge intensive, and you do not get locked out of content over lag.

If there is faulty "insert X" (netcode, hardware, data center placement, etc.) in one, it's not going to guarantee there is faulty "insert X" in the other.

I have no idea what you are asking, but the JP are having problems,too.

1.) Agreed. Hardware differences can impact latency, but from every route parse and ICMP poll, the latency happens before the hosts internal network.
2.) Alot of people blame the netcode. Though I have seen in (rare) instances that the transition from server cluster to server cluster is messed up (3102, 90000, 10105) it seems much better.
3.) The PS2 can support and play XI, but traffic is traffic. It's all the same internet (save for an extra hop through the PS servers which the PS2 didn't have)
4.) I don't think any MMO in general is data intensive. It's not about the amount of data that is being transmitted, it's about the the stability of the connection. Stable/low latency is much more important that up/down ratios
5.) Data transmission is the same. I'm pretty sure both XI and XIV use ICMP packets (pretty sure XI doesn't use TCP/UDP I could be wrong though)
6.) I think the reason XI seemed more stable is because they didn't have multiple data centers employed over continents. Their servers were located in the HQ (or close to it) so they could jump on an issue quickly without having to wait for the mothership approval (like Montreal). It's easier to manage one data center that doesn't have to chain and relay.
7.) I remember a little bit of rubber banding, but it doesn't stand out in my mind (XIV STANDS OUT)
8.) Agreed
9.) They don't employ the same server clusters for XIV and XI. Would be too much of a hassle to find points of failure. Separate the clusters completely and you can completely rule out a possible failure mechanism.

Imagine you wanted to go from your house (client) to let's say Chicago (host). Imagine that every home had it's own airport and airplane (client traffic). Your airplane is little so you can't fly there to Chicago on your own. This means you have to fly to your nearest commercial airport and connect flights (hub) where other people are doing the same thing (traffic). The hub isn't all that big so you have to get to LAX to reach your destination (node) where a TON more people are trying to do the same thing. Imagine that it's Thanksgiving. You can imagine that flights would be booked to the point there are delays (latency) as there are only so many seats the planes have. Now, imagine Delta (Backbone/ISP) lost your luggage or overbooked the flight and say you gotta wait (packet loss) because there are so many people trying to get places and they don't care because hey, they got their money. They know the system is broken, but are doing nothing to fix it.

Imagine the map and lines as flight paths. Imagine this kind of thing happening every single hour of every single day and you would have the internet.

Japan is an island. They don't need as much raw material, maintenance, monitoring, and repairs as there isn't as much ground to cover and the hubs/nodes are substantially larger due to land mass. But when you try to force more data than the nodes can route (which can and WILL happen) problems arise. I think this is why we don't see as many reports of latency from Japan.

It seems the main problem isn't on Square's end, but on the backbones and the ISP (Ormuco/Level3/TATA) that services hubs in the areas. They just can't support and route the sheer amount of traffic that is being sent.

I'm confident of this because I make a living as an IT admin for a company that aggregates, stores, parses, and graphs data from monitoring gateways in the field that report said data into our servers so it can be viewed though websites (either ours or an API call out). Oh well, I'll probably just be flamed anyway for trying to offer an explanation.

[Billie21]

So you're saying it's a problem with the area where the datacenter is located?

[Marishi-Ten]

So you're saying it's a problem with the area where the datacenter is located?

More or less. You can test this yourself by pinging your servers IP address. Famfrit's IP is 199.91.189.39 - http://arrstatus.com/

Open your command prompt and type (without quotations) "tracert 199.91.189.39" and hit enter. Watch for asterisks and high latency values. You will have to run this multiple times as the bandwidth ebbs and flows. You should be able to tell if the issue is on Square (asterisks are after Montreal) or on the node/ISP/backbone (before the Montreal hop).

[MythToken]

More or less.

So the location of the data server is the problem.... who placed the data servers where they did?

Also allyra is bringing up valid points, and not over reacting.
SE can't fix the internets problems which do exist, but they can optimize and that's all we are asking for.
This is what the beta should have been used for, and they should have planned their server location better.

kewl?

[Tupsi]

SE can't fix the internets problems which do exist, but they can optimize and that's all we are asking for.

Which is what they did and it still didn't fix some people's issues, but it also helped others that used to have issues - What this means is, SE can only do so much when it's other companies having the actual problem.

[Marishi-Ten]

So the location of the data server is the problem.... who placed the data servers where they did?

Also allyra is bringing up valid points, and not over reacting.
SE can't fix the internets problems which do exist, but they can optimize and that's all we are asking for.
This is what the beta should have been used for, and they should have planned their server location better.

kewl?

No. Physical location has nothing to do with it. Your route path and what nodes your traffic is flowing through is the problem. Put the data center in Middle Earth and as long as I have a stable route path or direct drop, I'm good to go.



Square bought the Montreal offices where their servers are located for NA/EU (formerly the Eidos offices). It is cheaper to maintain existing servers than buy new ones (price commercial grade servers and then triple that number for the things you need to interlace them and the logistics behind it. The number will will floor you). Square is a business. A business that is publicly traded and financials are scrutinized. You don't just drill 8-9 million dollars on servers and provider agreements for issues that are not your own for a fluctuating player base (servers that sit offline are counter to a return on investment and look bad to the board and the CFO. Modem businesses operate on a "justify being employed here" mentality. Once the player population stabilizes we may see server expansions and further fiber trenching). Again, this isn't the case as the connection issues come well before ORMUCO MONTREAL (I had dropped packets in Seattle) so for Square to be able to justify spending that much money to their majority shareholders is virtually non existent.

I'm not calling out Allyra or anyone specifically. I'm trying to provide reasoning, logic, and education (from an IT admin standpoint not a consumer) to the general forums. I'm all about pointing fingers and holding companies accountable. We just need to make sure we are pointing to the right people and getting things done from the floor up.

[Allyra]

FFXI stuff

I wasn't listing those to say that was the exact problems, just pointing out the differences to show why it's silly to say "if it doesn't happen in FFXI it can't be SE's fault for FFXIV."

As to the FFXI lag, I dunno how bad it was honestly, because the game didn't need you to have a perfect connection to play. I was just pointing out it definitely existed because of the running. But nothing in the game was going to kill you if you had rubberbanding. As well, it is a lot easier to notice in this game because of the red target circles letting us know exactly where things are going to hit, which FFXI doesn't have. So even if they were the same I don't think it would have stood out, or that anyone would even care.

I think this is why we don't see as many reports of latency from Japan.

We do see reports though, but I dunno how common they are. I have to rely on friends who can read the JP forums, but they do have it there.

It seems the main problem isn't on Square's end, but on the backbones and the ISP (Ormuco/Level3/TATA) that services hubs in the areas. They just can't support and route the sheer amount of traffic that is being sent.

But that would be on Square's end for putting all the data centers in one spot. Which is why I've been saying for awhile.

You can't honestly say that these hubs are able to handle WoW's traffic with millions of subscribers, but can't handle FFXI's traffic which is probably under a mil.

The difference would be Blizz has multiple datacenters so they aren't overloaded.

Again, not saying SE is 100% at fault, I never ever have. But they CAN do something about it. And some of us would like to know that they are actually doing something.

[Marishi-Ten]

snip

I see what you're saying, I do. But they way traffic routes and the way the entire structure is laid out wouldn't support the argument.

It's not Squares end. Their control starts and stops in Montreal (network wise). I'll show you (early hops omitted due to being behind my companies firewall):




Look at hop 6. I have an asterisk. This is command prompt saying that on the third poll, it could establish/verify the latency at that specific hop. Hop 6 happens to be in Seattle and is managed by Level3. Level3 is doing okay today, but has had issues in the past week: http://www.internetpulse.net/

That "data center" (it's actually a node) isn't managed by Square. Level3 is a Tier 1 backbone and are their own company that doesn't know that SE even exists. They are completely separate companies that do completely different things. It's unreasonable to expect Square to get on the phone to Level3 and say "hey, can you fix that node? Our players are having issues". Level3 would laugh them right off the phone and Square couldn't do anything about it.

Check hops 13-15. Now, these ARE Square's servers and internal route paths. If you see a spike in latency or asterisks here, then yeah, Square's bad.

This is just my route path. I'm one person out of MILLIONS that use the internet a day. Traffic ebbs and flows. There is literally nothing Square can do to fix a node that doesn't belong to them.

Find the break in the connection and run it back to the ISP that manages it. This is the only realistic way to fix the core problem.

WoW has better ROUTE paths. They probably have a contract with their ISP and full fiber connections to every switch. I'm sure Square asked for it. It's up to Ormuco (their ISP) to trench it.