| Author |
Topic |
CCP Fallout
 |
Posted - 2009.04.29 19:57:00 - [ 1]
We've recently made some very drastic changes to our network infrastructure. CCP Queeg500's newest dev blog gives us the scoop on the hardware and service provider changes we've made recently to help improve your connection to Tranquility. You can read all about it here. |
Akima Jarka
Universalis Imperium
Tactical Narcotics Team |
Posted - 2009.04.29 20:12:00 - [ 2]
Edited by: Akima Jarka on 29/04/2009 20:25:53
I didn't understand 1/2 that blog, but I think your saying something good happened.
So, Good Job!
Edit: Oh Great, I'm in a Fallout and Chribba sandwich....
Oh and Fir... ah screw it. |
Chribba
Otherworld Enterprises
Otherworld Empire |
Posted - 2009.04.29 20:14:00 - [ 3]
packet pron. |
Cors
It's A Trap
|
Posted - 2009.04.29 20:26:00 - [ 4]
I dont' know if anything was done "today" but I've noticed a significant increase in the speed of the www.eveonline.com website.
It used to hang for 10-30 seconds randomly for me. EVE always played fine, but the website was getting worse and worse to navigate through to the point I only browsed GD, CAOD(For ****s and giggles) and GDF forum to keep up with proposed changes.
Now it's fast.
If this was due to your recent hardware changes, Thank you! If it's a momentary warp speed of the website, and the lag will be back later.. DOH! |
Caldari Citizen4714 |
Posted - 2009.04.29 20:37:00 - [ 5]
This post has been cleared of inappropriate content.
Regards,
The EVE Online Moderation team |
Mal Plox
The xDEATHx Squadron
Legion of xXDEATHXx |
Posted - 2009.04.29 21:11:00 - [ 6]
Very nice! And a good job on implementation, i don't think anyone noticed any of the switch games you guys played! Good Job!
|
Arte
The Darkness Within |
Posted - 2009.04.29 21:45:00 - [ 7]
 Originally by: Caldari Citizen4714
TL;DR plaz.
I think it all means that your computer doesn't have to send a signal through so many routes to get to TQ and back. Ergo less internet based lag. Like the signal is a car that doesn't have to travel as far, can go on faster roads for more of the journey and if it meets roadworks, there are more options for diversions so that it can still get there quickly. Is that close anyone  |
Velicitia
Phenex Industries
PHOENIX. |
Posted - 2009.04.29 22:01:00 - [ 8]
pics or it's a lie  seriously though -- this sounds great. can't wait to hear more from the network team  |
Lakut
EmpiresMod |
Posted - 2009.04.29 22:03:00 - [ 9]
He misspelled "Deutsche Telekom".  |
Major Deviant |
Posted - 2009.04.29 22:24:00 - [ 10]
Nice. But...
Will this fix/affect the socket closed/client crash problem?
|
Dmian
Gallente
Gallenterrorisme |
Posted - 2009.04.29 22:29:00 - [ 11]
I'm not a technical person, but being around this "web thing" for so long (13 years!) I can have an idea of what's going on with all the packet routing. This post made me remember the days when Akamai appeared on the scene... Congrats, as always CCP is doing a great job. I'm really impress by the level of complexity behind this game. |
Arimathea Anthalas
Celestial Apocalypse
|
Posted - 2009.04.29 23:04:00 - [ 12]
It was me who made those comments, or at least I was one of the people making the comments.
I don't want you to have the impression I was against IOS-based platforms, despite what I may have written. My real concern is that you've moved to 76xx from 65xx which is really the same thing. We can argue the merits of the new supervisor, but while I think you have probably made a significant, substantial improvement, at prices customers normally pay, GSR (12k) is price competitive for your interface types I bet. And there are a number of options for CRS-1. Now you can also consider Nexus (for the switch side) and ASR1000 (for the router side).
Thanks for the dev blog, please continue to post these great detailed views on EVE internals and architecture, it is highly worthwhile to read. |
hubt |
Posted - 2009.04.29 23:08:00 - [ 13]
Did you consider using Akamai(they do more than just caching, they can improve routing too)? If so, why didn't you? |
Admiral Frools
The Illuminati.
Pandemic Legion |
Posted - 2009.04.29 23:17:00 - [ 14]
Excellent blog, most informative.
:thumbsup: |
John McCreedy
Caldari
Eve Defence Force |
Posted - 2009.04.29 23:25:00 - [ 15]
Nice blog, explained what you're doing in relatively lamens terms (which makes a nice change) but from the lamen perspective, whilst it explains what you're doing, it doesn't explain what you'll acomplish by doing what you're doing. |
Hoshi
Eviction.
|
Posted - 2009.04.29 23:40:00 - [ 16]
Originally by: John McCreedy
Nice blog, explained what you're doing in relatively lamens terms (which makes a nice change) but from the lamen perspective, whilst it explains what you're doing, it doesn't explain what you'll acomplish by doing what you're doing.
World Supremacy of course :) |
Semper Spero
IMR Corporation
|
Posted - 2009.04.29 23:43:00 - [ 17]
\o/ Packet Exchange has a node in Singapore. At the moment, I still see my packets being routed via L3 in the US before being routed via L3 in the UK and then to Package Exchange. About 300-350 ms round trip. I looking forward to see whether the latency improves after you turn on the 120 smaller ISPs! |
Jainia Soltella |
Posted - 2009.04.30 01:13:00 - [ 18]
Originally by: Caldari Citizen4714
TL;DR plaz.
CCP is trying to make the tube connecting your client to TQ shorter. |
MotherMoon
Huang Yinglong
|
Posted - 2009.04.30 03:09:00 - [ 19]
The internet is not a big truck |
Arch Alterius |
Posted - 2009.04.30 03:28:00 - [ 20]
can i has t2 networks? |
DigitalCommunist
November Corporation |
Posted - 2009.04.30 03:33:00 - [ 21]
I can't believe I understood that.  |
Aleyra Mel |
Posted - 2009.04.30 03:37:00 - [ 22]
Well taken from wikipedia: Quote:
Frequent misconceptions of the "tier hierarchy" include:
"Tier 1 networks are closer to the "center" of the Internet."
In reality, Tier 1 networks usually have only a small number of peers (typically only other Tier 1s and very large Tier 2s), while Tier 2 networks are motivated to peer with many other Tier 2 and end-user networks. Thus a Tier 2 network with good peering is frequently much "closer" to most end users or content than a Tier 1.
Cant see why you want so much to peer with a Tier 1 in order to be closer to the user.  |
Zex Maxwell
Caldari |
Posted - 2009.04.30 03:51:00 - [ 23]
I see why you have the server in London |
Maria Kalista
Amarr
Splinter Foundation |
Posted - 2009.04.30 05:44:00 - [ 24]
Quote:
We now have direct peering with: Global Crossing, Teleglobe, KPN, Telia, AboveNet and Level3.
W00t! |
Tribal Solidarity
Heavy Influence
Atropos. |
Posted - 2009.04.30 06:03:00 - [ 25]
One of the greatest blog posts ever.
The network team need to post more often and I WANT proof pictures of your sexy routing system! |
Charles Javeroux
Gallente
INTERSTELLAR CREDIT |
Posted - 2009.04.30 07:43:00 - [ 26]
yay...more speed to the little people |
CCP cNOC
|
Posted - 2009.04.30 10:37:00 - [ 27]
Edited by: CCP cNOC on 30/04/2009 10:38:50 Originally by: Arimathea Anthalas
.... Now you can also consider Nexus (for the switch side) and ASR1000 (for the router side).
....
I actually considered ASR1000 series in the beginning, but decided on going with the 7k6 as it's such a proven platform. We have had reports from ISP's having issues with the ASR. 7k6 are considerably more expensive than the ASR's, whereas they share some of the adapters. The 6k5 and 7k5 are now going separate paths, and they have improved the routing platform in many ways. There are several reasons why we keep the 6500 switches, but I'm not going into details why, but one major thing is we run the loadbalancers as a blades in the 6k5. About the wikipedia comment: "Cant see why you want so much to peer with a Tier 1 in order to be closer to the user" For us to peer directly with number of tier2 SP's will be hard, This is due to the fact that they usually peer with the T1's directly and we would have to run leased lines/fibre runs directly to them, and that is just not viable. So T1 is the next best thing. We though are going to use the PacketExchange peering platform that gives us direct peering with some of the T2's. The tiered model is not some grand scheme, just how things got organized as it's the viable way (IMO) There is another way to get closer to the end user ISP's, we are sitting in the same building as LINX. This is one of the largest "free" peering hub in the world, a meet-me point for many of the ISP's. This is something we are looking into and could be the next step for us. But I think the quality now of the Tier1 connections is that good, and we are mostly now seeing over congested lines from the end user ISP's to the Tier1's being the bottleneck. That unfortunately is out of our reach. |
Commander Azrael
Three Shades of Brown |
Posted - 2009.04.30 11:26:00 - [ 28]
Edited by: Commander Azrael on 30/04/2009 11:33:48Edited by: Commander Azrael on 30/04/2009 11:31:04Ooh, ooh we should directly peer so I get better response times while at work! Interesting blog, being a network engineer that actually has clients in telecity and just over the road in telehouse means I can find some of the finer technical points quite interesting. Presumably the load balancers in the 6500 switches are ACE cards? I've always wondered what the networking on the whole looks like on TQ and from these dev blogs I get to slowly piece it all together. Nice work guys Originally by: hubt
Did you consider using Akamai(they do more than just caching, they can improve routing too)? If so, why didn't you?
Akamai are more streaming content specialists, they're not really a backbone provider as such. The stuff thats going on here is internet backbone stuff, in an effort to reduce the number of hops people have to pass through before their traffic reaches the destination AS (TQ in this case). Limelight offer a similair service, and from some of the URL's I see CCP already use a lot of limelight services. The idea is to really allow some of the smaller ISP's to peer direct with CCP rather than bounce there conncetions through various tiered providers (such as global crossing, abovenet, L3 etc). Originally by: Tribal Solidarity
One of the greatest blog posts ever.
The network team need to post more often and I WANT proof pictures of your sexy routing system!
Take a look here. Should give the info on the 7600 routers, theyre quite big too at 6U high. But maybe it's just me that finds this geeky stuff intereting  |
Regat Kozovv
Caldari
Alcothology
|
Posted - 2009.04.30 11:45:00 - [ 29]
Originally by: Arimathea Anthalas
Thanks for the dev blog, please continue to post these great detailed views on EVE internals and architecture, it is highly worthwhile to read.
I agree. Keep it coming! Nice to hear from the players in the know as well, always makes for some pretty insightful discussions. =) |
Commander Azrael
Three Shades of Brown |
Posted - 2009.04.30 11:49:00 - [ 30]
Edited by: Commander Azrael on 30/04/2009 11:49:53 Originally by: Arimathea Anthalas
It was me who made those comments, or at least I was one of the people making the comments.
I don't want you to have the impression I was against IOS-based platforms, despite what I may have written. My real concern is that you've moved to 76xx from 65xx which is really the same thing. We can argue the merits of the new supervisor, but while I think you have probably made a significant, substantial improvement, at prices customers normally pay, GSR (12k) is price competitive for your interface types I bet. And there are a number of options for CRS-1. Now you can also consider Nexus (for the switch side) and ASR1000 (for the router side).
Thanks for the dev blog, please continue to post these great detailed views on EVE internals and architecture, it is highly worthwhile to read.
We've actually begin using Foundry XMR routers, for the price they're superb and so far theyre blowing the 72xx's we currently have out of the water (Not to mentioned looooads of interfaces  ). |
All Channels
EVE Information Portal
Screenshots & Art
