New on LowEndTalk? Please Register and read our Community Rules.
Interserver 6$ Storage VPS Review
I bought that just yesterday (in my timezone, I mean Aug.3), for 3$ actually because of that 50% coupon code.
It's actually pretty nice.
Here's the bench.sh
Here's the YABS, which everyone loves.
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## # # Yet-Another-Bench-Script # # v2022-06-11 # # https://github.com/masonr/yet-another-bench-script # # ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## # Wed Aug 3 17:29:29 UTC 2022 Basic System Information: --------------------------------- Uptime : 0 days, 7 hours, 39 minutes Processor : Intel Core Processor (Broadwell, IBRS) CPU cores : 1 @ 2397.222 MHz AES-NI : ✔ Enabled VM-x/AMD-V : ✔ Enabled RAM : 1.9 GiB Swap : 1003.0 MiB Disk : 984.2 GiB Distro : Ubuntu 20.04.4 LTS Kernel : 5.4.0-96-generic fio Disk Speed Tests (Mixed R/W 50/50): --------------------------------- Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ---- Read | 99.93 MB/s (24.9k) | 1.00 GB/s (15.7k) Write | 100.19 MB/s (25.0k) | 1.01 GB/s (15.8k) Total | 200.12 MB/s (50.0k) | 2.02 GB/s (31.5k) | | Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ---- Read | 2.84 GB/s (5.5k) | 3.43 GB/s (3.3k) Write | 2.99 GB/s (5.8k) | 3.66 GB/s (3.5k) Total | 5.84 GB/s (11.4k) | 7.10 GB/s (6.9k) iperf3 Network Speed Tests (IPv4): --------------------------------- Provider | Location (Link) | Send Speed | Recv Speed | | | Clouvider | London, UK (10G) | 449 Mbits/sec | 2.57 Gbits/sec Online.net | Paris, FR (10G) | busy | 2.59 Gbits/sec Hybula | The Netherlands (40G) | 672 Mbits/sec | 2.40 Gbits/sec Uztelecom | Tashkent, UZ (10G) | 904 Mbits/sec | 882 Mbits/sec Clouvider | NYC, NY, US (10G) | 3.64 Gbits/sec | 5.43 Gbits/sec Clouvider | Dallas, TX, US (10G) | busy | 4.42 Gbits/sec Clouvider | Los Angeles, CA, US (10G) | 423 Mbits/sec | 102 Mbits/sec iperf3 Network Speed Tests (IPv6): --------------------------------- Provider | Location (Link) | Send Speed | Recv Speed | | | Clouvider | London, UK (10G) | 1.25 Gbits/sec | 2.64 Gbits/sec Online.net | Paris, FR (10G) | busy | 219 Mbits/sec Hybula | The Netherlands (40G) | 1.05 Gbits/sec | 2.43 Gbits/sec Uztelecom | Tashkent, UZ (10G) | 1.13 Gbits/sec | 955 Mbits/sec Clouvider | NYC, NY, US (10G) | 6.68 Gbits/sec | 5.89 Gbits/sec Clouvider | Dallas, TX, US (10G) | 1.08 Gbits/sec | 4.50 Gbits/sec Clouvider | Los Angeles, CA, US (10G) | 1.21 Gbits/sec | 2.62 Gbits/sec Geekbench 5 Benchmark Test: --------------------------------- Test | Value | Single Core | 707 Multi Core | 704 Full Test | https://browser.geekbench.com/v5/cpu/16411553
Overall it's excellent. And I've never seen any HDDs that can run multi GB/s, so I would assume that this is some SATA SSDs or Many Many Many HDDs in a single RAID.
My aff: https://www.interserver.net/r/838364
pls buy through my aff, appreciate. OR https://www.interserver.net directly
My original blog post: https://frank-ruan.com/2022/08/04/interserver-6-monthly-storage-vps-benchmark-best-bang-for-the-buck-you-would-ever-see/
Oh, their IPv6 is kinda broken...
If they are running a mostly HDD SAN with SSD for caching, that could cause your performance tests to give results that are not representative long-term¹, the writes going to write-back SSD cache and the reads after coming from there also. It would be worth doing read tests on data that has been present and untouched for some time, at a later date when it is less likely to be in that sort of cache, to see what the performance is then².
 This is not a deliberate trick on their part, or a flaw in the benchmark overall, just the benchmark not being well suited to testing that sort of IO arrangement.
 Though note that as it is sold as a HDD based storage space optimised service, I'd not feel hard done to if the performance dropped off as I expect it will.
@FrankRuan is that a customized theme on WP? I can see elements of Neve. Looking nice!
Its a RAM caching. I have even higher speeds on my HDD-only server because of that.
Its good for you because recent files are in memory and good for provider as he can make cheaper server (perf of hdds can be somewhat slower and customers wont notice), but this is clearly not performance of HDD or RAID.
It can also be NVMe cache, but that still isnt HDD nor SATA SSD.
Im using similar package with them and only have good things to say about provider. Never needed a ticket or anything and performance is good
i thought this is a hdd with ssd cached too,
can we take advantage from this good IO for example to make storage vps as webserver? i mean for simple blog?
(i'm aware that a vm with pure ssd still fits better)
Any good way to do IOPS benchmarking bypassing the ssd cache?
Yes, I'm using Neve.
There's literally no non-aff links. They are not WHMCS, if you want to bypass my affiliate, then just visit interserver.net
You can just simply add
Affiliate link(s) must be clearly marked as an affiliate link with a non-affiliate link provided, in the same size. Affiliate links are a privilege for helpful, contributing community members. If you signed up just to spam, expect a ban. If your only posts are affiliate links, expect a ban. This includes posting links to your blog that only contains affiliate links, URL shorteners, etc.
Okay, I see.
But how on earth can I change that...
@Arkas (moderator) can likely fix that for you just remember in future to add non-aff link too
Looks like it's fixed @FrankRuan ?
Not reading the rules and fishing for affiliate credits from @interservermike - at least give the guy a mild warning
What a crime.
Or you swing the ban hammer
Oh thanks, I promise that I'll never do that again.
Don't worry about it man
I vote for @yoursunny to join the mod team because then the punishment would be at least making a push up video.
Thanks for the quick review. Plan to order one...
+1 for Interserver
Not in a VPS. Likely not on a dedicated box if using non-local storage. The cache is probably transparent to anything you can control so you can't turn it off or ask for your IO to bypass it.
With local storage there are OS options that can affect direct writes to final storage, but even with that it might not skip an SSD cache as that is non-volatile so considered final storage (they would skip a RAM based cache, or wait for it to flush to physical).
You can test bulk stream performance by writing something huge enough that the cache will start writing through instead of filling itself at the expense of other cached data - you'll see a performance drop at that point and after that you are measuring the next level more directly. For reading you can wait until all your data has been pushed from cache then it'll all come from the next level for the test, but you have no failproof way to know when the right time to test is.
Yes, I've used tar to pack a bunch of files up, it's like about 100MB/s, still acceptable imo.
Hi, everyone, I purchased one, now it has been 12 hours, but the detail page still shows: Power Status is: shut, what should I do? I use Debian 11 system.
Try power it on, if still not, send a ticket. They replies pretty fast.
ask support to install it, I had the sampe problem and their support installed it in no time.
Thanks, I asked staff in a previous ticket, but the response is so slow. I guess maybe I need to open a new ticket for this. Now I'm waiting for the responses.
I tried, it still not working, thanks, I've already opened a ticket.
@interservermike Any chance to restock in Los Angeles ??