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
This brings up another thought: I've been toying with the idea of having a few nodes served by their own SAN. A few 15k SAS drives in RAID 10 served as a ZFS file system over bonded gig-e. Haven't done much more research than that, though. Anyone else serving their LEB VPS's from a SAN?
I was being sarcastic, theres loads of "hosts" out there that claim "we dont use hard drives, we use cloud storage so theres no chance of your data going"
Don't worry, I got the joked and laughed
We mostly buy our drives new from NewEgg as well.
yes, what do you want to know?
How many nodes do you serve per SAN? I was thinking that it's possible that we could serve as many as possible til disk throughput starts to suffer, but it would seem like that would be too many eggs in one basket.
@Francisco @kujoe @mitgib: I know you guys have a whole pile of nodes. How often do you need to replace failed drives?
Not really a "pile", we only have 4 VPS nodes online with a handful more ready to be powered on with 3 non-VPS nodes running and so far we've only had 2 bad drives but they were DOA when they arrived (the 2nd bad drive was the replacement for the first bad drive).
/me knocks on wood.
I've replaced 2 failed drives this year, 1 of them wasn't really failed, but the node was over heating after moving it from SC to NC, and instead of a vent over it blasting 42F air at it, it lived in a cabinet, so things got wonky.
For the most part, all my drives are Western Digital RE2/3/4, Seagate Constelation, and Hitachi UltraStar, but now that I have smartmon tools 5.4x for EL5/6 I am going straight DeathStars, so I imagine failed drives might increase a bit, but the savings more than makes up for it.
+1 DeathStars in RAID w/ smartmon or HW raid card monitoring
+1 more as I use smartctl to disable deep sector recovery and HW raid
It all depends on the kind of nodes you have
As you know I'm new to the "low end market", last year I was not ever thinking it was possible to have nodes with the vps density you need to run this particular business.
For the lowend I use a Coraid SRX3200 that has 24 x 2TB disks
Partitioned with 1 10TB raid10 and 1 raid5.
and 2 x 10Gbps interfaces
This one is used by 10 servers in total including one big lowend node, and it run very well until now even if the I/O load produced by the lowend node is more than the sum of the other 7 servers.
The other low end nodes (4) run out of our old fiber channel infrastructure based on StorageTek 6300 with 14 + 14 sas 300GB disks (one raid5 for each node) and two indipendent controllers with 4 x 2Gb interface.
They have low peak speed but handle load very fine and you're able to maintain good latency.
Why did I use san in the first place? Well I had all in place and in particular for the StorageTek and old fc switches, they were sitting in the datacenter almost unused so it was an easy choice. They are doing a good job and I'm happy until now.
However the new lowend infrastructure (for kvm and/or xen) will not use SAN as it will cost too much if you have to buy new gears only for this....
So my experience is that SANs are a good thing if you (as we do) offer some high availability virtualization service as they allow you to do things like live migrations and node failover. Once you have in place then you can use them for other high available services as well. But the price is a big choice factor and as I told you I don't plan to buy SAN for budget services ;-)
Smartctl is nice. We usually don't get drive failures as we try to use new RE4 drives. We had a hard drive fail recently on one of our servers over 6 times in under 15 days.
Turns out it was simply a batch of WD drives that were DOA. :P
WD -1
With them acquiring hitachi, I guess it'll be zero-sum? :P
We have a pretty positive experience overall with WD RE4 drives, other than the recent DOA batch.
Samsung's are pretty good too.
My experience with the RE/RE2s has been nothing but great. It's the Blue/Green Desktop/Laptop drives that have let me down very often in the last 2.5 years or so.
Oh, and Seagate Barracuda's are pretty good too.
This is what was driving me, actually. I was thinking that, in the long run, it might be cheaper to buy a single array of X-number of SAS or enterprise-SATA drives and serve a few nodes (1-5 at least, or more depending on how much those 1-5 nodes load the SAN) over bonded ethernet. Since RAID 10 requires 4 drives at minimum, 5 nodes * 4 drives = 20 drives needed, if all 5 nodes have their own set of drives. Plus the power to run them.
I typically get mine from homeless people or WorldWar 1 Surplus Stores.
That's really not funny.
What a coincidence, neither are you
On a more serious note... do you serve any purpose at all here other than trolling or attempts to belittle actual users?
@taipres -- why not start your own "talk" forum? cheapvpstalk.net is available... I'm sure you'll make it way more high-end than LET
IMHO fiber channel is not an option unless you have some deal with some company that is decommissioning some infrastructure (you have to add the cost of HBA, FC switches and licenses and sfp to the san costs).
iSCSI never convinced me under heavy load until now.
You can consider Coraid and AOE technology as they are very cost effective
http://www.coraid.com/products/etherdrive_srx_series
You can use regular switches and ethernet cards (you do need some dedicated ethernet HBA if you plan to use vmware).
done
JUST ignore him :-|
I like eBuyer.
Their prices aren't that bad. http://www.ebuyer.com/173804-samsung-hd103sj-spinpoint-f3-1tb-hard-drive-sataii-7200rpm-32mb-cache-hd103sj
This guy with his horrible site mocking the entire VPS industry, then he comes on here making fun of homeless people...as I said, he's not funny, at all.
Do you just come here to get insulted by every single thing someone says about something not regarding you?
Why are you here again? oh yeah for attention, here's a thought, why don't you make buyvmtalk.com and actually provide real support to your customers, instead of ignoring them so you can act like some big wig on here? You're not a big fish in a small pond regardless of what you think and the fact a few fanboys on here kiss up to you in hopes for a free VPS.
No but I stand up for what's right, and if you find that guy funny you have issues. It's real simple, and mocking those who are struggling is sick and disgusting. The entire world is facing financial issues, and anyone can lose their job or face trouble in business at any time. That's NOT funny, period.
I take it you're not a Gottfried fan, either?
Why don't you make cheapvpstalk.com instead of getting mud on the carpets here?
Who's being ignored? We handle our support just as promised... so why don't you try to troll another provider who has threads opened about them? Oh, that's right, showing the same criticism towards anyone else would mean you'd have to stop showing bias. It's patently obvious to everyone here that you just think you're some big shot for 'standing up to BuyVM'... it just amuses the hell out of me that you tend to ignore everyone else that tells you to sod off. Must be living in one hell of a fantasy world
See, yet ANOTHER example of you not knowing what you're talking about. Anyone that's paid even slight attention would know that we don't do handouts, period. Hence, this begging you posit doesn't happen :P
Y'know, at least @Maounique would move on to a new topic every now and again. You're just getting boring in your predictability
@Aldryic You know Aldryic, I think he just fancies you.