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
@Nick_A Yes, if they're buying it specificly for that as a substitute to backups. Which is what my other point was referring to.
Marketing garbage that increases redundancy and IO which is essential for VPSs. I wouldn't want to trash that but then again coming from you nothing new there.
@Patrick See @concerto49's comment and then read "already" in my comment. Feel free to nitpick some more.
That is literally the stupidest thing I have heard anyone say on here so far...
@AnthonySmith wording may not of been the best, but same point behind it.
Let's word it a little more indepth, for the people that decide to be awkward.
Buying from a provider who advertises RAID 10, and buying a product for the sake of a RAID 10 array is stupid, redundancy and performance can both be questioned in RAID 10, or any RAID array for that matter.
There could be the same batch of drives and the likelyness of them drives dieing together is likely if it's a faulty batch.
Performance is the same as anything, for sure you can have providers with amazing performance in RAID 10 and then you have the providers with poor performance.
I'm not talking about any provider in specific, I'm speaking in general.
If you're saying RAID10 in and of itself is stupid, then you're of course wrong. If you're saying people should not substitute RAID10 for offsite backups, then you're correct. I feel like you're positioned somewhere in between those ideas.
@Jacob raid 10 is more redundant than raid 1 and 5, and it out performs raid 1, 5 and 6
If it is not set up great then maybe it wont perform as well as it could, perhaps that is stupid but raid 10 is not.
More IOPS, in raid 10.
With Raid 1 you can get up to 2 x read speed and no extra write performance, raid 1 can sustain 1 drive failure
With Raid 10 you can get up to 4x read and 2x write speed gain, raid 10 can sustain up to 2 drive failures
Raid 10 gives more IOPS than raid 1,5 and 6 and frankly in an environment where disk resources O/I R+W are shared raid 1 is just a poor man's version.
You have made statements about what could happen and putting down raid 10 due to random possibilities like bad batches which would affect raid 1 and 5 even more, you have backed nothing up with numbers or facts.
Hence me saying what you blarted out was stupid
I'm sorry, wheres that? Your also another butthurt UK kid
Chris talked to me on IRC, we have no problems so mind your own business.
You do realize that when you take a hotswap drive out of its bay it 'removes' the 'cables'. Idk how tight a fit everyone has with their drives in their cases and neither do you. I'm replying to your statement that everyone that doesn't use hotswap is stupid. (I use hotswap.)
It's not stupid. It means they've put forth reasonable investment. I'll take a provider advertising RAID10 on a quality controller before one advertising software RAID1. The reason is simple. It takes more to wipe out a RAID10 array than it does a RAID1. It takes more operations to demolish disk IO on two drives than one. That's why I advertise RAID10 and the controller I use. It's a buffer between two clients. Sure one client can use that whole buffer and make a problem, but it takes more physical action to get there, that's the point.
The distance from A to B is absolutely a relevant factor. Increasing that distance is not an irrelevant factor.
No one is calling RAID10 a backup solution. No one. The only people upset by RAID10 standard are people who are trying to cut corners and don't like that people have expectations out of their budget. In that case increase the budget. If you're going to do something, do it right. I know what it's like to cut corners. I did it. I'll sit right here and tell you that my first actions with CatalystVPS were poor. I'm not proud of it. I know I'm better than the route I chose to start.
@Patrick
Lol, I'm not even going to waste my time. You're collecting little points from my comments and then commenting on something miles away from what I said.
This went from RAID 10 is marketing garbage > the specifics of RAID 10 > The differences of RAID Levels.
What happened to the old LET...
@Patrick is butt hurt. I've got stock of new Mobos, RAM, Drives, CPUs.
The only thing used in my servers is the chassis, I have happy DS customers aswell as nodes running solid.
The fact that I also just brought a couple brand new supermicro barebones defeats @patrick's knowledge of what hardware I use.
I'm not wasting anymore time replying to this thread, nitpick all you want, you obviously have something against me which is pretty clear.
Yes from ebay, you blame your DC for failed drives when you use used HDDs. Excellent.
(Nothing wrong with used HW except HDDs IMO)
Have fun when parts die when you either buy from your DC for emergency replacement or have to ship them out with delay.
I could buy loads of HW, doesn't mean anything.
What does that have to do with anything? I've yet to receive negative comments on IO/Downtime/Network and other stuff so we're clearly doing well.
You offered 2GB LEB, you couldn't get any sales either then. Thanks.
Anyways, any user with HF history is always going to be a sore loser.
/Awaits 'boot'.
@Patrick One of the drives was damaged due to the tech dropping the server when being unracked, but amazingly drives have these things called "warranty", strange huh!
I'm not even going to bother arguing on the hardware, pretty sure I know my own stock.
I'm sure a DC wouldn't probably charge you for a hotswap drive replacement (especially if you already have replacement drives in cages ready - alas no screwdriver needed).
But if you don't have hotswap drives, and your server sits somewhere in the top of the rack, the tech needs to de-rack it (=remove network and power cables from behind too), open it, unscrew the failed drive and replace it, close the case and re-rack it. Far more complicated = a far more possibilities for the tech to screw things up PLUS you're definitely paying a nice remote hands bill.
I definitely don't consider my clients or myself stupid for buying/building HW-RAID10 nodes.
Performance and redundancy between RAID10 and RAID1 CAN'T be questioned. It's a fact. RAID10 performs better and is safer than RAID1.
No one buys a RAID10 based VPS because they think they can rely on it without taking any backups. Remote, off-site, backups should always be taken, and whether a provider offers them or not, the end-user should always take their own too.
Raid is not substitute for backups; however in seeing a provider make use of RAID it lets me know the chance of relying on those backups is less than a provider not using RAID. I then look at the RAID level implemented to give me an idea of performance/redundancy.
I think anyone not advertising their RAID setup when advertising is missing a trick and leaving a few "what ifs" for the customer.
Patrick and Jacob need to get a room
Not really, is that why we allow other game servers/vncs? You can probably run 4-5 HL servers, still kill IO but not as much as MC (Oh and customers do, read the LEB post - people utilise full 3GB). I care about my customers and disk latency etc unfortunately. They are also prone to DDoS, which you would know a lot about.
But if you don't have hotswap drives, and your server sits somewhere in the top of the rack, the tech needs to de-rack it (=remove network and power cables from behind too), open it, unscrew the failed drive and replace it, close the case and re-rack it. Far more complicated = a far more possibilities for the tech to screw things up PLUS you're definitely paying a nice remote hands bill.
There is a screwdriver needed in all the hotswap cases I have. You still have to screw the drive to the caddy, but I agree - if you aren't going to be doing the drive replacements yourself you will probably save some money and hassle by going hotswap.
Still doesn't make you stupid for not choosing hotswap though.
You can have replacement drives ready in their cages (or caddies if you will) so the tech only has to remove one drive and insert the other.
Of course it doesn't, especially if you're doing the replacements yourself
I was mostly referring to people how colocate in remote locations and need remote hands.
Isn't the answer to the question really depending on the packages you provide?
We currently offer Raid1 on our VPS. Why? Because we had a great node offering Raid10 15k SAS drives at a low price and nobody want it.
We had a test node offering low space, low memory and Raid1 hard drives and people couldnt get enough of it. On these nodes our stats might not be great, but we do not have people complaining about not being able to do what they want to do on these nodes and therefore we are happy, because our customers are happy.
How important it is to have RAID 10 set up on your VPS?
- Its importance depends on the product you are offering, your target market and the promises you make.
oh snap.... shit just got real.
exit 1
+1. This is a great way of putting this.
@httpzoom Good point. The real issue I have is that I frequently see people expecting to load up a high capacity OpenVZ node with RAID1 or software RAID10 with bad drives. It's not cutting corners if it matches your product, but if you're wanting to do what has always been considered fairly typical around here (load a node with OpenVZ, getting as much as you can in it without crippling it), you make a mistake by using sata drives and not HW RAID10. Having to undersell because you hit IO caps before the CPU is bad. Underselling because it fits your product or business plan, not bad.
@jarland,
Exactly. I think people around here are moving away from the way LET used to be. Providers used to give their opinions on things and weren't battling for business against each other.
This now seems like point scoring in most threads.
I don't see a benefit to scoring points. Should be obvious to all of us that at most around 10 potential clients care about threads that aren't offers here
@Jarland saying that I was a dick to someone in another post because they are simply a troll. I actually got 4 sign ups directly from the comment on the thread.
Luckily it wasn't another provider, but it was a person who seems to want to just be an annoyance / trash random providers.
I honestly think there should be a private section of this forum where providers can bitch between themselves. To often do we have "Would it be good to do...." I did it in the past when learning this market and looking back the advice I got was shitty! Everyone ends up finding their own way.
That depends on how you define "backup". A backup can be simply, "a copy of data on another physical storage medium." RAID > 0 achieves this.
So if I'm setting up a box with two disks, it's better to run on a single disk and use the second one for backups -- rather that installing two disks in RAID 1? That's silly IMO. Again, need to define "backup".
I didn't know that backups were inherently fireproof. Oh, you mean off-site backups
There are many different "types" of backups. A backup can:
Saying RAID isn't a backup isn't quite true. It serves one purpose of a backup quite well. But it doesn't provide redundancy and it isn't off-site
@httpzoom Nice. Well I think it's often mistaken here that providers are always speaking as providers. In some cases it's true, and my post about RAID10 is obviously from the perspective of a provider who regretted trying to cut financial corners in the beginning, directed to young providers trying to cut all the corners they can because they're not confident in their investment.
But sometimes I really do speak as me, not Catalyst
@jarland, thats the way it should be. If I was wanting to do PR for HTTPzoom I'd be doing a pretty shitty job.