Originally by: Emily Spankratchet

---

More nodes! Huzzah!

---

Originally by: Abdalion

---

Originally by: Jebidus Skari

---

What, in EVE, is a Tyrant?

---

Me. Especially when it comes to troll threads.

---

Originally by: Caol

---

Don't get too wasted on that "team building" exercise

---

Originally by: Jim McGregor

---

Edited by: Jim McGregor on 11/10/2006 18:35:02


Nice blog... but im still abit worried. Are you sure you can support the growing number of players by just adding more and more nodes? I keep saying that real time dynamic load balancing must be a good solution, but maybe its too complex and difficult to put in?

And Kali coming to the test server next week sounds fantastic. I will be there to help find the bugs if you do some fresh TQ mirroring. My character on the test server sucks. :)

---

Originally by: Grimpak

---

more nodes yay \o/

one might ask tho: extra number of nodes will not mean such great increase of performance that we might be lead onto?

I still belive that most lag issues are software related and not hardware-related.

---

Originally by: Skull Digger

---

Remember as a developer myself its always the hardwares fault. If you loose threads you cant blame the software.

I remember when there where hardware additions before and at the same time they changed the solarsystem jump gates. Seems CCP likes to do things that cant be traced to a specific cause but an over all mess.

---

Originally by: tbow10

---

isn't that somthing they teach you early in programing (101) not to change to much at one time take smaller steps in code so you know where somthing goes wrong when it does :P

---

Originally by: Oveur

---

[...]Not only that, we're always ordering something new and we don't always mention everything - but it's days like today which are fun to mention, because we have some big batches of hardware coming from IBM over the next months, resulting in more than 30 servers being added to Tranquility (that's 60 more nodes).[...]

---

Originally by: kieron

---

Snakes on a plane? F*(# that, we have Hamsters on Hardware!

---

Originally by: Miss Overlord

---

i think they went with intel on there but hve AMD on other parts of the beast

---

Originally by: Oveur

---

OK, I can take a hint. The new Tranquility consists of 70 x IBM LS20 AMD Opteron Blades, each Blade sporting two 2.4 GHz 64-bit Opteron processors with 2-4 GB of RAM, depending on application. That's a 140 CPUs, a full replacement of the current 140 Intel XEON 2.8 GHz CPUs.

---

Originally by: Bagehi

---

Just curious if CCP is planning on using some of those servers to hold the load from major fleet battles. I can think of a specific one that occurred over the weekend, and caused the node to crash repeatedly. And some frustrated GMs getting on and trying to keep the poor thing alive while players had their fun. Anyway... is it even possible to move a section of a node to a different server or is that physically impossible? I can only imagine the amount of data that would have to be shifted...

---

Originally by: Oveur

---

And indeed, there will be beer

---

Originally by: "Oveur"

---

To sum it up, in Samuel L. Jackson's immortal words:

"We're on the m***********"

---

Originally by: Pardack

---

Can we get some spiffeh technical details on what the cluster will contain after all of the mentioned node additions, RAMSAN and networking gear? Like how many RAMSANs are currently installed, total capacity, network pipe, unofficial place on top500 and any other geek-a-licious info you can come up with.

One question I've got is, which CPUs are being used in the LS20s? Is it dual-core, dual-cpu or dual-core and dual-cpu Opterons? Seems like there's many CPU options for them. From what I've inferred, it appears to be dual CPU/single core, but haven't seen it confirmed anywhere for sure.

---

Originally by: Annatar

---

I think we (the community) have lost a little overview on what Hardware (serverside) we are playing.

New Pictures included into a new Yaardware Blog would be sooo

Sweeeet.

---

Originally by: Magunus

---

Originally by: Pardack

---

Can we get some spiffeh technical details on what the cluster will contain after all of the mentioned node additions, RAMSAN and networking gear? Like how many RAMSANs are currently installed, total capacity, network pipe, unofficial place on top500 and any other geek-a-licious info you can come up with.

One question I've got is, which CPUs are being used in the LS20s? Is it dual-core, dual-cpu or dual-core and dual-cpu Opterons? Seems like there's many CPU options for them. From what I've inferred, it appears to be dual CPU/single core, but haven't seen it confirmed anywhere for sure.

---

If I'm reading this right, they're dual CPU dual core 64 bit opterons. Towards the end he mentions 70 LS20s and 140 CPUs. He also mentions that the CPUs are 2.4 GHz, and if you follow the link to the servers, and look at features and benefits, the only way to get 2.4 GHz is to go with dual core, single cores are 2.8 GHz.

---

Originally by: Pardack

---

Originally by: Magunus

---

Originally by: Pardack

---

Can we get some spiffeh technical details on what the cluster will contain after all of the mentioned node additions, RAMSAN and networking gear? Like how many RAMSANs are currently installed, total capacity, network pipe, unofficial place on top500 and any other geek-a-licious info you can come up with.

One question I've got is, which CPUs are being used in the LS20s? Is it dual-core, dual-cpu or dual-core and dual-cpu Opterons? Seems like there's many CPU options for them. From what I've inferred, it appears to be dual CPU/single core, but haven't seen it confirmed anywhere for sure.

---

If I'm reading this right, they're dual CPU dual core 64 bit opterons. Towards the end he mentions 70 LS20s and 140 CPUs. He also mentions that the CPUs are 2.4 GHz, and if you follow the link to the servers, and look at features and benefits, the only way to get 2.4 GHz is to go with dual core, single cores are 2.8 GHz.

---

I saw that as well, but note the "up to 2.40GHz" and "up to 2.80GHz", these are maximums. They very well could be running single-cores @ 2.4GHz.

---

Originally by: Kuolematon

---

Originally by: "Oveur"

---

To sum it up, in Samuel L. Jackson's immortal words:

"We're on the m***********"

---

Did you ask permission for this via eve-mail?

---

Quote:

---

Nertzius > having fun being incompetitent?

---

Originally by: arjun

---

ccp has set their priorities constantly wrong for the last 3 years. we cry about fleetbattle-lag since 3 years. it never changed.
my guess is, that there is a fundamental flaw in the gamecode and nobody dared to touch it because it would require a rewrite of the whole thing. so we have seen constant doctorings and half arsed "improvements".
why on earth are they refusing to give us the option to remove the graphics if wanted? fleetbattle with overview only - why not? they just can allow themself to refuse without giving a acceptable reason since 3 years.

---

Originally by: voodoo

---

Well new nodes more servers thats great considering you're bringing 8 new regions into play with kali.

---

Originally by: kieron

---

Texas Memory Systems sent the new RAMSAN out today and is on its way to the server inLondon, UK for installation. Depending on when UK Customs clears it, installation will most likely be within the next 2 weeks.

---

Originally by: Cyleth

---

Edited by: Cyleth on 13/10/2006 16:14:19
People will always complain 'cause of something. The problem atm, however, is not lag. It's the dieing nodes and it's quite annoying. There's no way to fight big battles without a)having a node death, b)having lag so much that you cant see a thing within 10mins, c)simply can not control your client at all and see how you get popped.

I seriously do hope that that extra yaaarrrrrrdddware is going to give a significant boost on server and we actually can have decent territorial battles.

Such news as Oveur's blog is always good news.

---

Originally by: Cyleth

---

The problem atm, however, is not lag

---

Originally by: Cyleth

---

b)having lag so much that you cant see a thing within 10mins.

---

Originally by: Pardack

---

Adding more systems/regions will simply spread out the problems a bit and not fix the problem. You've now got more systems to load balance and admittedly more resources available, but isn't this just again 'bandaiding' the problem? It appears now that with the addition of more regions/systems and with the addition of more 'yaardware', that performance gain will basically be a wash?

---

Originally by: Pardack

---

IMHO, more resources should be focused on the majority of 'laggy' systems and WHY they are so laggy. Yes, POS wars suck AND cause lag, but is that not putting your focus on a single area of performance and not addressing the other areas of EVE that despite these changes, will still experience the same lag (eg. non 0.0 players)? Unless that is casuing 90% of the lag.

---

Originally by: Matthew

---

More systems doesn't automatically mean significantly more load. Empty systems will be generating very little load. If players go to the new systems and generate load, that means that they are not where they would otherwise be. Just opening up 8 new regions does not mean we will magically get 8 new regions worth of new players to match. Therefore, the new regions may lead to a transfer of load from other places, but it shouldn't generate more load.

---

Originally by: Matthew

---

Bear in mind that the node a laggy POS war is on, is likely to also be the same node that other entire constellations are on too. The POS war will make all those constellations laggy too, because they're on the same node. Eliminate the POS war lag, and you immediately un-lag all those other constellations too.

---

Originally by: Matthew

---

With the second RAMSAN in, would be interesting to know how much of the DB can now be fitted on the RAMSAN rather than slower disk arrays.

---

Originally by: Pardack

---

I semi agree here. I didn't mean to come across as saying that it will automatically generate more load, empty systems don't consume many resources. Having said that, with the constant influx of new players and the expected 'rush' to the new regions to claim them, they will most likely have more load than average, especially when they are first rolled out. This also means that any other systems that are sharing those nodes will experience more load than average. The number of nodes being added is more than the number of systems being added resulting in more systems overall sharing a node.

---

Originally by: Pardack

---

Agreed, while I may have been concentrating on the majority of players performance, the key thing is reducing the most lag-causing problems in EVE. Then again, since most players are concentrated in empire space, if something else is causing lag for most of them, doesn't it make more sense to fix that problem first?

---

Originally by: Pardack

---

Granted, a laggy POS war will affect other systems. Perhaps separating 0.0, 0.1>0.4 and 0.5+ into 'mini clusters' for load-balancing may help alleviate the problem some. Since most 0.0 systems are unused there would be more resources for POS wars and low-sec/high-sec wouldn't be affected and vice-versa. Jita-like systems wouldn't have an effect on 0.0/low-sec systems. Unfortunately this would also most likely not be the most efficient usage of resources. I'm not one for band-aid fixes, but it seems worthwhile to look into. A few nodes would go mostly unused, but it would also serve as a 'reserve' for high loads as some have requested, which is not currently possible with the current structure of EVE. Call it a workaround.

---

Originally by: Matthew

---

Adding new regions is unlikely to significantly increase the number of new players coming in, so that factor would be the same either way. If people are flocking to the new systems, sure, the load is higher in those new systems, but it will be correspondingly lower in the systems those players have left. A couple of days load-balancer shuffling, and everything's happy again.

---

Originally by: Matthew

---

We're adding 8 regions and 60 nodes. Now, I'm not sure exactly how many systems those 8 regions have. But lets look at the node expansion. We know from the previous yardware blog already linked that there are 140 nodes. Therefore 60 nodes is 43% of current capacity. Therefore, if the 8 new regions are going to maintain the same systems-per-node ratio as currently, they would need to contain as many systems as 43% of the currently accessible eve universe. Given that there are currently 33 regions listed on the alliance territory map (quickest source I could find), and that's only 0.0 regions, the 8 new regions would have to be ridiculously huge to contain enough systems to maintain the current systems-per-node ratio.

---

Originally by: Matthew

---

Sure, it's best to target what's hurting the majority of players first. But with the structure of the cluster, empire lag could just as easily be a side-effect of 0.0 fleet battles as a seperate problem of it's own. Especially when nodes start dropping and systems have to be moved to new nodes.

I would think the load-balancer does this to some extent already. The problem is that with the current number of nodes, to maintain performane in high-sec you would end up with huge swathes of 0.0 on the same node and a fleet battle would have the potential to take down massive areas. All those new nodes will give them more headroom generally, which will let them have more "spare" capacity balanced into low-sec areas which see volatile traffic levels, ready to cover fleet battles, because they'll have the capacity to do it without removing required power to meet more predictable loads in empire.

---

Originally by: Pardack

---

Edited by: Pardack on 14/10/2006 03:50:18

Originally by: Cyleth

---

Edited by: Cyleth on 13/10/2006 16:14:19
People will always complain 'cause of something. The problem atm, however, is not lag. It's the dieing nodes and it's quite annoying. There's no way to fight big battles without a)having a node death, b)having lag so much that you cant see a thing within 10mins, c)simply can not control your client at all and see how you get popped.

I seriously do hope that that extra yaaarrrrrrdddware is going to give a significant boost on server and we actually can have decent territorial battles.

Such news as Oveur's blog is always good news.

---

You contradict (sp?) yourself. You state:

Originally by: Cyleth

---

The problem atm, however, is not lag

---

and then you say

Originally by: Cyleth

---

b)having lag so much that you cant see a thing within 10mins.

---

Lag IS the problem. Whether it's caused by BMs, multiple fleets jumping in on each other or "Carebears" farming missions, lag is the result.

Dynamic load balancing is the key IMHO. No, it's not easy with the current game structure, but it is at LEAST semi-doable and semi-feasible.

---

Originally by: Viktor Fyretracker

---

i know its fleet zergs causing node drops but for the bookmarks why not move them clientside? its not like anyone stands to gain anything if they did hack them. i mean they cant give their raven 50,000 shields 98% to all resists and 1.0 refire with torpedos.

---

Originally by: Blacklight

---

Normally I'd applaud the addition of extra hardware, hotfixes, future code changes to improve performance but it's very difficult to be upbeat about the future despite this good news.

The current state of the game is depressing and after nearly four years of it I find it increasingly difficult to keep telling myself it'll get better when it doesn't, it gets worse.

Sorry to rain on the parade but for a great many of us involved in large scale combat (hell, not even large scale these days to be honest) the game is dying a long, slow and painful death.

---

Originally by: Hygelac

---

Firstly I'd like to say how deeply appreciated these blogs are, most of us from other MMO communities will understand how unique CCP is in itÆs responsiveness and general communications with the customer i.e. Us. :)

Now hereÆs the bad bit. In over 2 years of playing Eve, I think I can safely say that IÆve never seen the game in a worse state than it is currently, and itÆs at the current point where even the hardcore cadre of optimists are starting to loose faith in CCPÆs ability to actually pull this off.

I think CCPÆs shining moment in terms of hardware and performance was the introduction of the new cluster. We all marvelled at the smoothness of the change-over, we all wowÆed in corp chat about how smooth the whole game was and lag-free the fights were. Sure by the time you got to 100v100 fights things were still a bit laggy, but even those fights were noticeably improved. Before long people were deploying 150 man fleets against one another and I even saw the odd 200 man blob in combat à a testament to CCPÆs hard work and the organizational abilities of the player base.

When was that? March? April à I canÆt remember the specific date. But I can safely say than since that point Eve has been progressively worse and worse and even worse. I think itÆs safe to say that I lived through some pretty bad days pre-RMR/Ramsan but even those bad times are like a momentary blip compared to the current poor state of things:

IÆve started to see threads and strategies on how to deal with ænode-crashesÆ and how to create node crashes, and drastic measures by the DevÆs which seem to make little difference at all. Things are at the point where anything more than 50 on 50 is risking a node crash and when the node comes back, the basic mechanics of the game, even those goals people were fighting over in the first place were bugged/unstable until the next down time.

WeÆve seen systems loosing sovereignty, POSes shooting friendlies/not shooting hostiles, IÆve even seen station settings no longer working allowing whole fleets of enemies access to assets they have previously lost. The worst point for me is where things have gotten so bad that GMs have had to intervene in normal player activities and have even influenced the course of combats, all in an effort to try and keep the servers up.

Eve has never been better than it is today, but at the same time, itÆs never been more unplayable and that is frustrating us (and no doubt you guys too) to the point of tears. I think I speak for the majority of the player base when I express my admiration and love for this game, When even the die-hards are starting to loose faith, there is a serious problem: ôThis cannot continue.ö

If it requires a slow down in new content/functionality deployment then so be it. Shiny toys are great, but what use are the toys when we canÆt play with them. WeÆre with you Oveur, please make Eve great again, whatever it takes!

Signed Eve Fan-boi #413,514,165

---

Originally by: General Meridus

---

Yes, it gets slow with 200+ in a 0.0 system, but its manageable.

---

Originally by: Pardack

---

Edited by: Pardack on 15/10/2006 19:19:29

Client side BMs, while nice are very open to hacks. No they won't give 50k shields and 98% resist, but they would be able to instantly warp to any player/structure (POS) which IMO is the crux of the problem with client-side BMs.

Not to mention being able to create "uber" safe-spots in any system.

---

Originally by: Gunther Dwendel

---

Concerning "lag" casue by the physics of the game. Perhaps CCP should investigate the feasiblity of the discrete physics processor or "PPU" (Physics Processing Unit). Developed by a new startup company called Ageia.

---