Linode Compute Dedicated vs Shared CPU Performance Benchmarks

[@joshhaas7240]

Hi Egee! I actually found your ASMR channel first, and was pleasantly surprised to learn you do tech stuff over here. I'm a developer at Linode, so noticing this video made me do a quick double take (small world, no?). You're correct that the "g" stands for "generation", but in this case we're using it to refer to plan generation. Sometimes Linode does free upgrades (albeit the last one was a while ago now; 2018) so you get better specs (e.g. more disk space) for the same money. Some of our older customers are still running grandfathered-in configurations, like "g4-standard-2" or some such. Upgrading requires a reboot and a migration, so sometimes people just don't do them. That being said, it looks like we don't have this documented anywhere, so I bugged our docs team about adding it n_n; Personally I use dedicated for minecraft servers, which in my experience makes a big difference for modded. Anecdotally I most commonly see people use them for databases. In theory shared plan performance depends on hardware + steal, whereas dedicated only depends on hardware. If your shared plan didn't land on a particularly busy hypervisor, I guess I'll take that as a win haha. Definitely agree with "only pay more if you get a benefit from it". Albeit I know some customers like the peace of mind knowing with dedicated there's zero possibility of steal appearing in the future, despite possibly no immediate performance increase. P.S. I can't tell you how happy I am you pronounced Ubuntu correctly; it boggles my mind how many giant channels don't! P.P.S. This is the second time I tried posting this comment; I'm assuming I just had some dumb network issue ... fingers crossed I'm not spamming you with a duplicate.

[@Egeexyz]

That's all super cool to hear!! I just moved my personal infra from Vultr to Linode so I plan to do lots more benchmarks like these pretty soon.

[@joshhaas7240]

@@Egeexyz Oh neat! Do you mind sharing what prompted the move?

[@Egeexyz]

Sure! - I actually switched *from* Linode to Vultr a couple years ago because their instances were way faster at the time and I was looking for the best value. The speeds are comparable now and I prefer the developer docs & tooling that Linode offers over Vultr's.

[@pz8424]

Hi, I'm the dev behind helsing. It's amazing to see people using it ^^

[@Egeexyz]

That's great!! Thank you for making it

[@tHe0nLyNeXuS]

Interesting content. I haven't checked out the benchmarks you use, but the 7642 is a 2nd gen processor, and has more cores and a way bigger cache than the 7601. That might explain the difference in performance, if the workload fits into cache. Simply saying that the 7601 is "inferior" is misleading, though, in my opinion (albeit a generally fair statement). If you had a workload (or multiple loads) that would nearly saturate a 7601, you would probably get better performance out of that dedicated older CPU than the shared 7642.

[@jay-ar6593]

So is linode's CPU model always an AMD EPYC #### @2.1Gz? If it is that seems rather slow to me compared to what other companies offer in terms of CPU speed.

[@CarlosPedroche]

Hello, what others company do you recommend for Dedicated CPU at the same price ($30) of Linode?

[@gabboman92]

do tests on contabo too. I am using that one, theyre cheap but single core performance is.... not the best. but lots of threads!

[@Egeexyz]

I've never heard of Contabo, thanks for telling me about them! I plan to look at as many clouds as I can (not just the big ones) so I'll definitely add them to my list!