Storage safety, RAID?

edited August 2014 in Features and Ideas
Hi, Does VULTR backup data automatically? In case of disk failure, will all data be lost? Does VULTR use RAID for storage? Thanks.

Comments

  • edited August 2014
    I can't answer for VULTR, but you generally need to sort out some such solution for yourself on such an unmanaged service.

    I use http://www.tarsnap.com/ for my backups - they are reliable, secure, and cheap.
  • We do not, but you can take snapshots for backup. We plan to offer a mission critical backup service in the near future.
  • For the record, "We do not" is in reference to backing up data automatically.
  • I may have missed this, but do you guys use RAID? Wouldn't want to get my VPS reset in the case of disk failure...
  • I'd assumed the original answer meant 'no RAID', but now it's ambiguous!
  • @BensDaMan

    From their Twitter, Storage Servers are in RAID 10 configuration with SSD acceleration.
  • edited October 2014
    Thanks @John315!
  • Hosts must have some sort of RAID for their entire volumes (not just single discs). However - the information about RAID is a bit short.
  • As someone who is not a 'celebrity', and doesn't have an overinflated sense of the shite I spout, and has no interest in 'celebrities', I don't have - and will never have - a twitter account.

    I hope Vultr aren't making any twitter-only announcements.
  • @BensDaMan - note he just mentioned the NJ SATA storage systems
  • edited October 2014
    @jamie I see that, luckily my SATA VPS is the only one with anything on it that i would care if lost. Of course I still want to know about their SSD plans....
    About the Twitter case, I'd fully agree, they should list this everywhere, not just Twitter. Though you don't need to have a Twitter account to see their feed! ;-)
    http://twitter.com/TheVultr

    @tram12 This is true, but what if it was RAID 0? (I'd definitely assume it wouldn't be, but thats an example.)
  • edited October 2014
    @BensDaMan Don't come back with sensible answers! This is the internet!

    Seriously though, cheers!
  • I'm just sad that twitter axed their RSS availability for feeds. :(
  • I'm interested now....is there RAID on all plans? (I hope so)
  • I have tried to ask what redundancy has been put in place by vultr to prevent issues but all my discussions keep being closed :(

    I don't think that asking a question about redundancy should be closed & it should be openly discussed...

    Anyone know what they use & why vultr won't even answer a simple question?

    https://discuss.vultr.com/discussion/473/storage-node-redundancy-info-required
    https://discuss.vultr.com/discussion/474/why-was-my-discussion-closed-without-an-answer
  • I'm not trying to be rude, but gleaning over existing forum posts tells me the following:

    - There is no redundancy or automatic fail-over, just raw speed.
    - You may choose to include automatic backups on your VPS for a minor extra fee
    - Said automatic backups are not stored on the node -- they are stored elsewhere, so if the node goes down, the backup is safe.

    Having said that, all this means is you should simply treat your VPS instances as any infrastructure should be treated - monitored, backed up, checked for consistency and with a disaster recovery plan.

    Vultr has some of the cheapest prices and best performance I've ever seen, having a mass storage node to do backups or paying a nominal fee for vultr to do it seems trivial.
  • @Arffeh - thank you for your answer & yes I did try to find the info in the forum & that's how I found this thread which had no official answer from vultr which left other people asking similar questions.

    If your answer is correct then it would of just been easier if vultr answered my simple questions instead of just closing my discussions.

    I am not trying to put vultr down as they obviously have a great product (except for the current sydney issues) but it sounds like they are just selling vm's which shouldn't really be marketed as "cloud servers".
  • Now now guys why would Vultr reveal all what good is it if you can't hide some competitive secrets.

    Regarding raid - they don't get them sort of benchmarks without you know some sort of performance based raid hope that gives you some sort of insight.

    The good ol fashion "cloud servers" does come about from time to time but remember cloud is not only about high availability after all Softlayer markets there hourly bare metal as cloud - that's not to say the definition of cloud has got lost over the years.
  • Vultr doesn't have any security Raid1 or 5 o 6 or 10 in its servers ? Vultr, why don't you answer this questions ? How come?
  • @keyjey - looks like your thread was closed just for asking a simple question too: https://discuss.vultr.com/discussion/479/raid

    @respite - our questions are not asking for any competitive secrets & are answered by all other hosting companies (most the other companies even advertise how redundant they are!)... We have the right to ask what we are buying, we have the right to ask how redundant the servers are, we have the right to ask how protected our data is.

    We would simply like to know if we are buying real "cloud servers" as advertised (or are we just be buying vm's)?

    Do these severs have automatic fall over to another node if the node they are on fails?

    Is the storage located on a raid protected san?

    To me, the only reason vultr will not answer our questions (& close all new threads about this topic) is because they are trying to hide something.. My guess is that they are not selling real cloud servers & that they are just selling vm's (which is ok for the price if you just want vm's & not real cloud servers - but we would like to know the truth/facts about what we are buying).
  • I was going to start the "SANs are in fact lower availability compared I with a local raid solution" But I'd rather state your asking the wrong questions the question you should be asking is what is the overall "uptime guarantee" associated with factors out of my control such as hardware, network and power... you can then base your buying decision in the hands of that answer.



  • edited January 2015
    Sorry "raid protected san" should of been "redundant raid protected san" ;)

    This is how real cloud hosting should work (from a competitors website - minus the competitors name): http://i.imgur.com/E0TAp68.png

    Traditional Providers
    -Use hard disks that are physically located inside each host that provides virtual servers.
    -Store VPS data on one physical host, creating a single point of failure.
    -Require an outage to transfer a virtual server to a different host.
    -Have I/O performance and disk capacity limited by the host hardware.

    Real Cloud Servers
    -Use SSDs that are connected to virtual servers via a high-speed storage network.
    -Store data on multiple hosts, avoiding interruption if an individual host fails.
    -Can perform "live migrations" to transfer a virtual server to a different host without an outage.
    -Can increase I/O performance and disk capacity by installing additional storage hosts.

    I am just trying to work out if vultr is more of a traditional provider or they sell real cloud servers..
  • This is a deal breaker for me.. I am not interested in a tradition provider (otherwise I would just buy dedicated servers) - I want a real cloud provider as thats what I thought I was buying. Which one are you vultr?
  • Hello - sorry to bump an old thread - but I am very keen to find out the answer about whether the hardware is redundant. I could not find any suitable answer about this - so for now I conclude it is not.... and therefore if the node goes down, that means my websites will be offline until fixed or ??
  • edited June 2017
    I have no connection to Vultr, so this reply isn't authorative, but I hope it is accurate.

    "You pays your money, you makes your choice"

    Yes, if the node goes down, your site wil be offline until it's fixed. Similarly, if a disk goes pop, it's possible you'll lose data. These are sold (very cheaply) as unmanaged VM servers.

    I don't remember where, but this information is available - I remember researching it before I came onboard.

    Having said that, it's not "abandon all hope" - things work well, and I've personally found Vultr very fast at addressing issues. After all, customers wouldn't hang around for long if the reliabilty and support was rubbish.

    But It's ultimately up to you to sort out things like redunancy, backups, load balancing, fail-over, cdn etc.etc.etc.

    As you can see, vultr do actually offer some of the above as options, but otherwise you're on your own.

    As an aside, vultr *may* use raid in some circumstances, or other redundancy features - simply as a way to reduce headaches on their side of things - it's up to them. The point is, the VM's are not sold with such guarantees, and as such my guess (and it is just a guess) is that this is why they don't want to discuss what setup they may literally have internally - just the service they offer us.

    In conclusion, I like the flexibility of cheap nodes, and I can sort out all the other bits myself. If this isn't your cup of tea, I guess more managed and expensive solutions exist elsewhere.

    But then, I'm a paranoid sort of person who has to do everything myself to be sure :-)

    Cheers!
  • @jamie - good and fair answer ;-) and it pays to be paranoid where data is concerned :wink:
  • @ClapperTrap Thanks! Don't get me wrong, I've had my own share of disasters over the years (not on vultr)

    I was once so paranoid, I had a bare metal server a few years ago (again, this was pre-vultr) - i had another one that I used just for remote backups, and I paid a 3rd party backup company too... Sorted! ?

    I was unfortunately in hospital for a while - my credit card expired, and was automatically replaced, but each company needed the card details re-entered.

    By the time I was out, *ALL 3* had gone, through account-deletion. I lost years of work, including every bit of facebook code I'd written, emails, the lot.

    Separate hosts. Separate locations. Separate companies..... But still, single point of failure... payment!

    Maybe I'm not paranoid enough!
Sign In or Register to comment.