New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
@jsg could you please share with us ping & traceroute for the american locations used in your speedtest?
For the SSD vs SAS debate I did some testing (not on netcup) and in many cases providers will limit iops or the node will be busy limiting the difference with a dedicated HDD, but of course the result is often better than on a shared HDD. As netcup seem to use an SSD cache on their SAS nodes, the difference could be pretty low between both system for most workloads. As you said, if you organize your setup taking RAM into account (cache, DB index in RAM, redis DB for stuff that changes often with some optimization) a faster drive isn't necessarily needed.
By the way, on many SSD VPS, CPU can be the bottleneck way before i/o. And there is no need for super fast i/o if you can only use two slow cores as (for example) your mysqld will use all the CPU and your drive will stay far from being used at its full capacity!
SSDs especially in environments used to virtualize servers probably get much more customers crammed onto their nodes nowadays because it somewhat opens up the IO bottleneck. if done wrong you might see other ressource shortages instead then.
so of course SAS might give more value with less customers on the same node or even the same performance as with ssd if hooked up and shared in a reasonable way. on top fs cache for servers can work so much different from your usual windows experience...
comparing drives in virtualized environments to personal use cases on desktop PCs and calling it all wrong based on that is just narrow minded.
Short answer:
It depends.
Here you go:
Also note the about 4 ms to FRA.
Exactly. In a shared environment you are more likely to hit quickly a bottleneck, be it i/o or something else. SSD doesn't mean that storage is gonna be fast as hell. Just that it's less likely that i/o will be the bottleneck, or if it is, that it's not going to happen as fast as it would have with a spinning disk. And that's where netcup seems pretty good, their "root" nodes look well balanced, maybe thanks to the "dedicated threads".
Thanks!