| Author |
Thread Statistics | Show CCP posts - 13 post(s) |
|

CCP Fallout

|
Posted - 2009.02.16 15:45:00 -
[1]
In the words of the infamous Mindstar... "that's what she said!"
Speaking of Mindstar, he's just published a new blog on what's coming with Thursday's extended downtime.
So what is coming? A brand new RamSan. What does that mean exactly? The death of database lag... shorter expansion deployments... shorter morning downtimes! And, most importantly, a very excited and happy Mindstar!
Fallout Associate Community Manager CCP Hf, EVE Online
|
|
|

CCP Mindstar

|
Posted - 2009.02.16 16:01:00 -
[2]
Oh yes we have plans to go to 2008 Windows / SQL also..... all in good time  -- |
|
|

CCP Claw

|
Posted - 2009.02.16 16:29:00 -
[3]
Originally by: Rhys Onasi CCP Mindstar overcompensated a bit too much (for the potential dullness of this dev blog) with his cheeriness... I think he's spent a little too much time alone with the server... 
He wasn't alone <3
|
|
|

CCP Fallout

|
Posted - 2009.02.16 18:04:00 -
[4]
Originally by: Katabrok First
Originally by: CCP Fallout In the words of the infamous Mindstar... "that's what she said!"
Speaking of Mindstar, he's just published a new blog on what's coming with Thursday's extended downtime.
So what is coming? A brand new RamSan. What does that mean exactly? The death of database lag... shorter expansion deployments... shorter morning downtimes! And, most importantly, a very excited and happy Mindstar!
Ok man, will it be thursday or wednesday? In the news linked in your post, it says wednesday, not thursday.
Thanks
It's Wednesday. I got the day wrong initially. Sorry for any confusion.
Fallout Associate Community Manager CCP Hf, EVE Online
|
|
|

CCP Mindstar

|
Posted - 2009.02.16 20:40:00 -
[5]
Originally by: N'olive
Quote: In total, we now have 32 of these ultra powerful blades in the cluster - 26 running as SOL blades, and 8 running as Proxy blades.
26+8 = 34, not 32 ;)
I'm good at servers not addition  -- |
|
|

CCP Mindstar

|
Posted - 2009.02.16 20:49:00 -
[6]
Originally by: Altomat How much performance improvement do you except for the Nehalem Blades over the Wolfdales? If they perform nicely, will the remaining AMD blades upgraded to Nehalem ones?
Altomat
That is actually a big unknown for us at this point. We have had various experts in the field claiming anything between a 5% and 100%+ performance increase.
One of the things about the EVE server is that it is essentially a single-threaded app at this point in time. The Wolfdales are very good for EVE as they are simply very high clock speeds. The newer processors coming out from AMD and Intel tend to focus more on multiple cores than on high clock speed. That is great for multi-threaded applications, but not so much for EVE.
The Nehalems do have a few interesting tricks up their sleeve - like turning off all but 1 core and then overclocking the one that is still running - which may give us quite a boost. Until we get some real world tests in, we don't really have any solid idea.
The current AMDs within the cluster are quite fine for their task, but they will eventually need replacing at some point. When we need to replace them, we will have really good knowledge through testing of the best processors to put in. -- |
|
|

CCP Mindstar

|
Posted - 2009.02.17 14:53:00 -
[7]
Originally by: Ephidaurus Anyway, I get the feeling that the Mainframe part is not really understood. It is still in line with CCPs idea of having one big cluster (even CPU cluster) that hosts the virtual world. It only has better I/O capabilites then an "ordinary" blade center can offer. And it is within one frame^^. But that is true with a blade center as well ;-). It is true that it might not be as fast as a blade center. But I understood that their problem is I/O and not "plain" cpu power.
Our problem is actually both I/O *and* raw processing power, but in different areas of the cluster. These two are pretty independent of each other for our purposes, and the architecture of the EVE server works rather well.
In our database, we have the tremendous I/O requirement that is satisfied quite nicely by our RamSans, but our database is not a CPU monster by any means, and nor does it need to be at this time.
The blade servers are what provide our raw cpu power for running the game simulation, and the application can be fairly evenly distributed over any number of these, so at scaling the clusters processing power is fairly easy to do as well with the approach we currently use. Best of all they are easy to switch out if and when we need to (eg replacing half the cluster over 2 regular downtimes). -- |
|
|

CCP Mindstar

|
Posted - 2009.02.17 15:41:00 -
[8]
Originally by: Mashie Saldana
Originally by: Yayoi Tan In the terms of databases, may I ask how many SQL queries per second/hour EVE takes :) I would believe this is an astronomic number LOL but a bit curious.
In a devpost 1-2 years ago they mentioned 155 million queries a day which is just over 1800/s average.
During peak hours we do 2000 - 2500 transactions per second. Around 40,000 disk IO operations per second. -- |
|
|

CCP Mindstar

|
Posted - 2009.02.17 17:11:00 -
[9]
Originally by: Spathi Why retire any hardware (its not that old)? Just re-purpose them for other things.
That is precisely what we have done. Retired from Tranquility, but not from use. -- |
|
|

CCP Mindstar

|
Posted - 2009.02.17 18:54:00 -
[10]
Originally by: Reptzo
Originally by: CCP Mindstar
Originally by: Spathi Why retire any hardware (its not that old)? Just re-purpose them for other things.
That is precisely what we have done. Retired from Tranquility, but not from use.
out of curiosity, what do you do with the old hardware? add it to sisi? Would just setting the old hardware to only run the statistically less used/empty systems be better than migrating all systems to the faster hardware? or making the old hardware do the "bt1ch" work help?
(some but limited topic knowledge)
Some of them went to a much needed expansion of the test server network, others got shipped back to Reykjavik to become internal servers just within CCP. We have plenty of need for servers outside of TQ, and in most cases the AMD's have more than enough power to do the job 
Why not keep the older servers in production? It would spread load a bit among the cluster, however even at 208 nodes we still have quite a bit of headroom on overall cluster CPU. Power usage within the datacentre is always a consideration, so any ability to consolidate servers without compromising performance is a big plus. -- |
|
|

CCP Claw

|
Posted - 2009.02.17 18:56:00 -
[11]
Originally by: MotherMoon
sisi?
For various different things actually - don't forget that these are still super awesome blades, just not quite as super awesome as the newer ones; there's always a use for lots of super awesome blades :)
|
|
|

CCP Mindstar

|
Posted - 2009.02.18 12:45:00 -
[12]
Originally by: Ki Sol I know flash drives are fast (internally to the raid) and doing backups and updates MAY be faster but if you are going to reduce your Input/Output from 400k to 100k and your bandwidth from 3GB to 2GB, How is this going to REDUCE lag?!
We are adding have added the RamSan500 alongside our two existing RamSan400's. The 500 is replacing a fiber channel disk array, and our usage of the 400s will remain unchanged.
Instead of using 2Tb of plain old hard disks, we now have 2Tb of Flash Raid  -- |
|
|

CCP Valar

|
Posted - 2009.02.19 12:52:00 -
[13]
Originally by: Naga Tokiba Did I read somewhere:
Shorter daily downtimes
But that dont seem to apply to today I guess 
We are still doing some changes. Today I was moving disks between disk arrays.
We currently have 6 enclosures, but only need 3 for backups and such after we moved the data to the RAMSAN 500. The array that houses the "quorum", the database for the Windows clustering service was partially on the enclosures we are removing, so I had to fix that this downtime.
Tomorrow, I'm moving the backups from the current array to the former primary data enclosure and after that we can turn off the 3 enclosures we are retiring.
---- Virtual World Database Administrator Operations department CCP Games |
|
| |
|