Did anyone else land on an E5-2630L in Fremont?

We have two Fremont linodes, both 512, just migrated to 1024. The first migrated in roughly 7 minutes, the second took a little over 17 minutes. No big deal, just a random observation. The thing I found odd was from /proc/cpuinfo on the first:

model name      : Intel(R) Xeon(R) CPU E5-2670 0 @ 2.60GHz

And on the second:

model name      : Intel(R) Xeon(R) CPU E5-2630L 0 @ 2.00GHz

From Linode's blog, "Powering our NextGen hosts are two Intel Sandy Bridge E5-2670 processors. The E5-2670 is at the high end of the power-price-performance ratio."

Did they forget to mention that this is not entirely true for some of you unlucky SOB's in Fremont? :-)

16 Replies

I've seen E5-2630L's in London and Atlanta. I found they performed worse than the old hardware so I moved to 2670's now they fly!

@obs:

…so I moved to 2670's now they fly!
Care to share how you did this? Is there a deterministic way other than luck of the draw?

@dffdce:

@obs:

…so I moved to 2670's now they fly!
Care to share how you did this? Is there a deterministic way other than luck of the draw?

You should be able to request a migration to 2670 hardware via a ticket if needed.

I was able to, and I did so because I was having issues with single threaded tasks on the 2630L.

The 2670 did indeed solve my problems though!

E5-2630Ls, among other processors, exist in all Linode locations. If you feel like you would benefit, feel free to open a support ticket and we can move you to a server with E5-2670s.

I've just upgraded a server in London and specifically asked (twice) for a 2670 since I know the 2630L's perform poorly. I was told the migration was configured so I migrated. Turns out I was dumped on a 2630L I asked Linode why, they said because they had no 2670's left. Why on earth would you say you can migrate to a 2670 when you don't have any.

I offered to upgrade to a larger node to see if there was a host with a 2670 for that plan size. I was informed there was one. I confirmed I wanted to migrate. Over half an hour later, no response.

At this point I give up. The server's sticking with the crap 2630L until it can be moved to another provider.

Linode you're support and service have degraded a lot lately. I miss the good old days when you were small and your support was second to none.

@obs:

I've just upgraded a server in London and specifically asked (twice) for a 2670 since I know the 2630L's perform poorly. I was told the migration was configured so I migrated. Turns out I was dumped on a 2630L I asked Linode why, they said because they had no 2670's left. Why on earth would you say you can migrate to a 2670 when you don't have any.

Linode you're support and service have degraded a lot lately. I miss the good old days when you were small and your support was second to none.

Agreed! my London 2048 was on a L5630 node before the upgrade and I was landed on L5520 after the upgrade. I asked politely to migrate to the new E5-2670 nodes and support replied with a generic response saying sorry for the issue and my migration has been configured.

I have did it and landed on 2630L even though I asked for 2670. And performance was horrible and I was seeing CPU steal 20% or more on any given time. I just went to a new provider. Didn't bothered to ask again.

@tasaro:

E5-2630Ls, among other processors, exist in all Linode locations. If you feel like you would benefit, feel free to open a support ticket and we can move you to a server with E5-2670s.
Any other suggestions? Opened a ticket but was quickly denied and then engulfed in "Not simple because of issues with newer versions of Xen…" etc.

I'm not ready to jump ship but so far this experience is VERY "un-linode like". It's starting to smell like tactics others use, you know, sling technical jargon at 'em until they acquiesce.

If you're in Fremont you're kinda stuffed since you're having a migration force on you. I purposely didn't migrate this server since I knew the 2630Ls were crap and I'm happy with the performance on the old hardware. However the old hardware had a problem so the migration was forced. The migration was scheduled for 11 days from now. I decided to go early on a Sunday morning when everything's quite on the server. Then I find out they didn't have any 2670s left. If they'd told me that in the first place I'd have waited until they had 2670s or the 11 days came up whichever came first.

OK after 2 hours of seemingly non sequitur machinations - finally got the second node on a E5-2670.

model name      : Intel(R) Xeon(R) CPU E5-2670 0 @ 2.60GHz

Don't know if it was simply 'cause I tried this ticket on a Sunday or what, but eventually we got there…

What's the deal with the nextgen hardware? I think it might be time for a bit more transparency regarding what you're doing - because customers are getting different levels of service.

> We’ve begun a refresh of 3/4 of our entire fleet to a new “NextGen” host hardware specification.

Is the E5-2670 the minimum CPU we can expect on servers? What are the remaining 1/4 of servers running? Is this referring to the fact that you've got to upgrade 3/4 of the servers, or that you intended to leave 1/4 of the servers on older hardware?

I assume these upgrades are still ongoing? What's the status of each datacentre for upgraded servers? Any timeframe estimate for remaining upgrades?

I'm currently on a L5520 @2.27GHz in Fremont and I'm about to upgrade for the RAM. If I get the E5-2630L will it be better or worse than the L5520?

I got lower Unixbench scores on the 2630Ls than on the L5520s. Not really sure how that translates to real world performance, but that's the benchmarking.

We had terrible performance on the 2630's. Support was not helpful, said there was no issue with the host. Something is very very wrong with those CPUs. Migrating away from one of the hosts took over an hour for a 1024 Linode. We asked support if there was anything they could do to speed it up. All we got back was "sorry, out of our control". Definitely ask to be migrated to a 2670. Maybe if they see enough hurt from the 2630's, they'll finally let us know what's really going on with those.

For our nodes the L5*'s were noticeably faster than any 2630 we landed on. Keep in mind, the 2650's do not seem to be plagued by the same terrible performance.

Ended up copying a Linode to another server to get off an E5-2630L, got an L5520 instead.

I hope that these servers performing badly are the exception and not something to expect as standard. I've currently got another 2 servers still on servers with the same CPU - both are still running fine, but I don't think I'll hesitate to switch if the performance drops on them.

I ended up moving to a different provider instead of trying to find a decent host server at Linode. It's a shame really Linode has some really good features but it just wasn't worth spending the time trying to find a decent host machine when we could migrate once to a different provider that we know performs well.

As of a short outage a few hours ago, I've been moved off the 2630 onto a 2670.

Neato.

Reply

Please enter an answer
Tips:

You can mention users to notify them: @username

You can use Markdown to format your question. For more examples see the Markdown Cheatsheet.

> I’m a blockquote.

I’m a blockquote.

[I'm a link] (https://www.google.com)

I'm a link

**I am bold** I am bold

*I am italicized* I am italicized

Community Code of Conduct