Why cloud services rock

Earlier this month I went away for a week. Two big things broke while I was gone:

In each case, software or hardware that I maintain went badly off the rails. I try pretty hard to lock down my WordPress. I protect my /wp-admin/ so that only certain IP addresses can access that directory. But this was a weird vulnerability. If I had hosted my blog with WordPress as a web service, I wouldn’t have had to patch things up myself. And I’m pretty sure that at least a few of my draft blog posts leaked to folks who exploited the security hole on my blog.

My NAS box was even more painful. I have a Buffalo TeraStation, and it’s worked well for over a year. But they don’t always cope well with power outages. And the version that I have required me to undo 22 screws to replace the bad hard drive. Then I had to wait for five nervous hours to see whether the NAS could repair my RAID array. No data was lost, but suffice it to say that I’ve got much better backups now, plus an uninterruptible power supply (UPS) as well. Again, if I had storage somewhere in the cloud, it would probably be more reliable than my ham-handed attempts at backing up my data.

So there you go: I went on vacation for a week and came back to two scary reminders why hosted services and storing data “in the cloud” (on someone else’s servers) can be better than doing it yourself. In general, a well-run cloud service is going to be much more reliable than any average person would be.

45 Responses to Why cloud services rock (Leave a comment)

  1. When one considers how cheap and powerful flash drives are – just get the most gigs you can get – and do automatic backups to that as well.

    It is not uncommon to see 12 to 64 GB drives now

  2. What kind of “cloud services” do you use or suggest to other people?

  3. Sorry about your loss.

    My focus this weekend was also uptime, not loosing data, and having things work.

    The thread on this took wings of its own over at Webmaster World.

    http://www.webmasterworld.com/supporters/3549640.htm

    I have been using I-Drive the last few weeks which backs up my laptop via wifi. It took a little getting used to, but is working good now.

    It will be great when all of our files can be online. There are still many times that it is hard to connect. In Mammoth last week, I just could not get a connection, no matter what I did, not even at Starbucks. On the other hand, on the entire road to Vegas last month I was able to stay connected via my blackberry.

    I had the most amazing experience at Apple tonight. It was actually the first time ever playing with Apple computers with the idea I might actually buy one. I had never thought of them as “real” computers. What is going on there is mind blowing.

    The only reason I might wait is the upcoming release of Googputers ๐Ÿ™‚

    Matt, last year your list of what events you were going to was very helpful. I discovered Wordcamp which was world class, and I never would have come across without your schedule. Perhaps you would like to publish your list again this year. I am already making plans for Defcon ๐Ÿ™‚

  4. I couldn’t agree more. A few years ago I was running dedicated hosting machines in two data centers to maintain the domains for my family’s computing needs on hutchison.org. Today almost all of that is hosted in the cloud. It’s freed up a great deal of my time, and allowed me to relax more when I’m traveling. More information on the services I use at:

    http://allen.hutchison.org/2008/01/cloud-services-rock.html

  5. Exactly, and not just storage. With EC2 you can have Amazon like scalability, add on top of that WeoCeo for load balancing and you truly have a win win with the front end. However, now they have SimpleDB in beta. One thing I hate the most about web dev is scalability issues with the database. This is the year of Cloud computing for sure. My projects will be taking full advantage. Not only do you get all the pros of doing such, but damm it’s cheaper than managed services.

  6. That is until the cloud goes down.

    ๐Ÿ˜‰

  7. I am surprised that you did not already have a UPS on your big drive. I of course learned that lesson the hard way myself, twice. Once was a power outtage and then the power surge that followed when the power returned, causing my main computer to fry several components. The next time was a nearby lighting strike that fried most electronics in the house, including 3 computers, which the damage apparent spread through the cable modem, router, hub and through the cat 5 cables which were not run through a UPS. All are now though!

    Thanks for the additional info about further locking down wordpress btw.

  8. Unless its with Google!

  9. It is for that exact reason that I have been using Amazon S3 + JungleDisk.
    Takes a while to upload all your data on ADSL, but once it is up there you have piece of mind that your data is backed up redundantly by a reputable company. The plus (paid) version of JungleDisk even allows you to only upload the changed parts of files. I would recommend it, although other similar services are of course available, e.g. XDrive.

  10. I would love to be able to store my backups on the “cloud”, but I have nearly 50G of photos growing at around 1G per month. At the moment it’s still far cheaper to buy disks than pay to back this up somewhere on the network.

    In my mind some form of P2P co-operative backups would be the best way round this, effectively share your backups round a number of people, I would donate 100G of space and get 25G of network backup?

  11. Hi Matt,

    It sucks when you return from vacation to find something bad has happened, and I’m sorry to learn of your WP draft exploits / power outage.

    What’s even worse is when you discover such things when you’re still on vacation, as I did when my domain name was stolen. It put a downer on the last week of my holiday in India.

    As you know, through one of my blog posts being Sphunn, my domain was returned, and I restored all the original content. Everything was great until last Friday, when all my pages have been dropped from the Google search results, so my traffic is down a huge amount (for my meagre standards).

    I’m not exactly sure why this happened, but I’m guessing that some dodgey content was displayed during the six weeks the hacker had control of davidairey.com.

    I’ve submitted a reconsideration request, with all the info I’ve gathered, and hopefully that can be looked at before too long.

    All the very best.

  12. Chris, perhaps the problem will be solved soon. I’m beta testing a hardware image storage solution that’s very promising. It analyzes images with pattern recognition software then uses select images and pieces of images as baselines for comparision to perform lossless compression on the remaining ones. As the number of images, especially of similar subjects, is increased the compression gets smarter/better. When I had around 75GB of images stored it took a little over 125GB of space on the storage device do to overhead but now I’ve got over 350GB of images stored and I’m only at 155GB of storage space used. The system estimates I can gain an additional 25% reduction if I reindex/recompress the entire image base using later additions as bases. No idea on eventual pricing or even how accurate these figures are since they are all reported through the (windows only) interface to the hardware. Pretty cool though.

  13. Omar Yesid Mariรฑo

    Yep, storing data by yourself is a nonsense when the prices for storing data โ€œin the cloudโ€ is now really inexpensive.

  14. I don’t even think the majority of geeks have the bandwidth to make this work. It sounds wonderful since making backups is still a bit tedious, even as it’s made simpler. If I needed to backup my entire hard drive (boot records, OS, media and all) I would need over 500 GB of backup space–and the bandwidth to upload it. After a hard drive crash I would then have to wait to download the 500 GB (and in what format) so I can restore my machine. It just doesn’t sounds reasonable.

    If you are using data on the cloud, where as it’s stored in the cloud and accessed in the cloud; that’s a different story. However there are still speed, efficiency and bandwidth issues here as well.

  15. Then the question is.. why don’t you use blogger? ๐Ÿ˜‰

    Yes, this is a (very light) flame.

  16. I’m a little surprised that you don’t use Blogger, really. Of course, it’s not without its own problems, but I generally found it less broken than WordPress. I ultimately ended up moving to Movable Type, which I find nicer than either, but Blogger was a decent second.

  17. I can’t believe I’m agreeing with SEW on something, but in this case he’s right. There are a lot of cheap flash drives available (here’s a 16GB for $99.99 CAD, and there are probably places in the States that carry them a lot cheaper because pricing on this stuff sucks here), and many of them do include at least some security measures.

    Back your data up, put the flash drive on a neck strap, and off you go.

  18. Hosted services have their own problems as well. Look at all the blogspot blogs that were down for hours when Blogger was down. The only Blogger blogs that were available were ones that were self hosted on external servers.

  19. Adam, and SEW

    You can get a mac time capsule for $500 – and that has a terabyte of network storage…. or I can get 80 gigs of flash drives…

    I know I won’t have to deal with hard drive failures, but for that price per gig and storage size, i’ll take my terabyte.

    Also, if you’re backing up multiple websites, versions of software, OS installs, or a lot of music, you can need that much space.

  20. I have always had a ups setup to protect against power outages, but I host all my servers at 365 main street in SF which has been pretty good to me over the last 6 years. BUT – even they have issues – there was a power outage which identified a problem with the transfer switch which brought down the whole data center effecting many large sites. My servers were only down for a few minutes but the damage was done and I spent a few hours getting them back. I consider 365 main to be a pretty good facility – a vision of Dave Rand of abovenet – but even that CAN have problems. I am considering a multi datacenter setup but that gets somewhat pricey. I am glad you got your data back because it would have been hard to do a recovery on that Buffalo raid.

    Keep posting stuff like this – I thinks its really interesting!

    J –

    John

  21. David Airey, I wouldn’t worry about it much at all. As recently as Dec. 22nd when Google visited the page we saw a parked domain (I believe that’s because it was transferring back to you), so you do expect to see a few issues as Google regains its trust in your domain. I see your site when I search for [www.davidairey.com], and I think it will return to ranking normally pretty soon.

  22. Boring Marketc, I do use Google Docs, Gmail, Google Calendar. I like that I can export my data from all of those web services. Lots of people like wordpress.com and Flickr and Smugmug and similar services too, of course.

  23. David Klein, so far it’s just SMX West in Santa Clara in February, but I’ll try to do a post if I’m doing more stuff.

    Chris Fleming, I’ve often thought that that would be a cool service.

    Robert Synnott, the plugins on WordPress really got me hooked. Lee, it’s true that cloud-based stuff does go down from time to time. But even then, they don’t often lose your data.

  24. John, will do. I was really worried I was going to have to install Ubuntu on my wife’s mac to mount the RAID in software mode to start recovering data. That would have been ugly.

  25. Thanks very much for getting back to me, Matt. That puts my mind at rest. Cheers.

  26. the problem is with a lot of these cloud apps there run on shoe strings look at all the sites that went out when 356 main had its “problems” and more recently 37signals.

    356 main was a disaster waiting to happen it wasnโ€™t a real UPS it was some Dutch half assed system apparently they hadn’t sprung for the dark start system nor did they test often enough.

    thereโ€™s also security and escrow problems if a supplier goes bust how do you get your precious baby pics back in 1/10/20/50/100 200 years time.

  27. thanks for the warning about your NAS, my routers / servers have a UPS, but I had assumed that there was little reason to add one to my NAS.

    – do you think it was a shutdown problem or a surge problem?

    @ Chris Fleming – have a look at Wuala:
    (google tech talk)
    http://www.youtube.com/watch?v=3xKZ4KGkQY8

    – can’t get it to build properly on Linux 64 bit though ๐Ÿ™

  28. Its true that “a well-run cloud service is going to be much more reliable than any average person would be”. The problem is that the average person has no way to check if a certain cloud service is well-run or not.

    Earlier this month Telia, the biggest telephone company in Sweden, lost all email to about 300000 customers that was delivered between 15 Dec 2007 and 3 Jan 2008. It turns out that they didn’t check the server logs of the backup server.

    This if from a company that have had a large ad campaign about the cloud storage solution: “store your image on our super secure server”. (Granted, the “super secure server” service may be run in a better way than the email server, but I still don’t trust them.)

    I’m having a hard time trusting a cloud service unless I know one of the admins personally (and sometimes not even then, because of the horror stories they tell me ๐Ÿ™‚

  29. Has anyone every tried the Amazon service for online storage and data bases? Currently I back everything up to various gmail accounts. ๐Ÿ˜‰

  30. Matt, I’ve been reading more and more about the partnership with IBM for cloud computing, from an article on msnbc about thinking like a googler. I think its a neat technology, but it spooks the heck out of my pop as he’s nervous about what’s his not being his anymore. Sorry about your NAS going down, but did you think of me when it did?
    http://www.mattcutts.com/blog/rules-of-thumb/#comment-28454

    heeh

  31. The problem with Cloud services is still the ISP. In Canada we have to pay extra for large amounts of bandwidth. Plus the speed is really appalling. For an individual user, it’s much faster and easier to set up local backups or use Apple’s Time Capsule on a mac. If I had to wait for my 100+ gigs to back up to a services over the internet, it might be next week before my backup finishes. Not to mention that it tops out at about 80KB/second at which point my download rate is also brought to a screeching halt. I pay the extra $25/month for unlimited bandwidth but what is the point with such slow speeds?

    The ease and convenience of a home network is still the way to go until the technology and services catch up with our needs. I once had a friend connect a car battery to his UPS. He extended the the uptime to hours and could wheel around his whole computer setup in the office. He’s a photographer so he needed to move the computer around the studio when shooting. I thought it was a cool way to provide some serious power backup at a cheap price.

  32. Cloud computing should spook more than youfoundjake’s dad. Well, maybe “spook” is a little strong, but there are genuine issues of privacy and of data ownership.

    For example, I have no control over all the stuff I create using Google’s great tools. Today, Google promises not to give that information to anyone, absent appropriate legal processes, but that is only for today. What if Google changes it’s policy? What if the laws change?

    Look at the gmail privacy policy. It specifically states, “Residual copies of deleted messages and accounts may take up to 60 days to be deleted from our active servers and may remain in our offline backup systems.”

    Before the .com bubble burst, there were a lot of companies that promised to keep information confidential. However, when they went belly up, the bankruptcy courts decided that their customer data was an asset the creditors were entitled to sell or whatever.. So much for the promises of confidentiality.

    I’m not saying Google is going belly up any time soon, but what about the more marginal cloud companies? This is where Google, as a market leader, has to do more than not do evil, it has to do good by pushing this issue and figuring out ways to protect privacy before it becomes a genuine problem. As Haldeman said, “Once the toothpaste is out of the tube, it’s awfully hard to put back in again.”

  33. I’m a big fan of using Amazon’s S3 for backup. I bought JungleDisk and I use that for backing up my photos (just photos for now, until I can clear my backup backlog). Unfortunately, on Comcast’s cable connection, I can only upload about 3 GB a day, so it is slow going. JungleDisk works well from my Macs and PCs.

    I love S3 because I don’t have to have the large upfront costs of multiple disks for redundancy (or lots and lots of DVDs) and the challenge of finding a secure offsite location to store my data (those external hard disks or optical media won’t do any good if your house burns down). Also, in theory, I can take advantage of Amazon’s scale: as they purchase new machines, the cost per storage goes down. Hopefully they’ll pass that on to the consumer at some point.

    Andrew

    Full disclosure: I’m currently employed by Amazon, but I was excited about S3 before I got the job.

  34. David Airey, you’ve had a stressful enough few weeks, so glad I could ease your mind.

    Andrew Hitchcock, is this the Andrew Hitchcock that posted the Google whiteboard bus pictures? That was great: http://andrewhitchcock.org/gallery/v/misc/googlebus/
    It genuinely looked as if they copied a Google whiteboard; we tend to scrawl and then let other people add on until it’s the markers are layered on quite a bit.

  35. Jake, I found myself saying “A NAS is not a backup. A NAS is not a backup.” over and over. ๐Ÿ™‚ I think stuff like Time Capsule is brilliant because no one is going to pay attention to backups until a) they lose data or b) it’s dead simple and probably on by default.

    Andrew, I love the idea of S3. But I can’t imagine why they haven’t added a mode that makes it look just like scp. That would let S3 slot right into peoples’ backup strategy.

  36. Yeah, that was me Matt ๐Ÿ™‚ I’ve posted here in the past, but not for a number of months.

    You might want to check out s3sync: http://s3sync.net/wiki It is supposed to be similar to rsync, but copies to S3.

  37. Greetings. As others have noted, there can indeed be significant potential privacy and security issues with cloud computing, however with sufficient implementation care many of these can be significantly reduced — though in some cases this might require the cloud provider to give up or at least limit some potential monetization vectors, which could be an obvious problem in a “free” services environment.

    But an equally serious issue for the rise of the clouds could well be the increasingly bitter battles over network neutrality, bandwidth, and other ISP-related controversies. While there are all sorts of optimization possibilities, cloud computing (by definition) depends on having sufficient affordable bandwidth (from both ends!) to support the application mix, and the current trend with ISPs is ever more restrictions and attempts at limiting user options, not encouraging bandwidth-rich applications.

    Yet another reason why the network neutrality debate is so important.

    –Lauren–
    Lauren Weinstein
    lauren@vortex.com

  38. Matt

    Sorry, if my post didn’t help with your question. I hope i didn’t violate any TOS with the comment i made. I thought it was a well thought out response to your question. Please, if you get a moment I would appreciate a response as to why you trimmed my response so I can ensure in the future they meet your specs.

    I really enjoy your blog and find it very informative and an interesting source of information. Again, my apologies if I did something not meeting with your expectations for this blog.

  39. See, you cannot go on vacation. The year before last we went on vacation in January only to come home and find the igniter on our 2 year old boiler (Heating System) had failed and the house was down to about 40 degrees. Thankfully no pipes froze up but that first night back the outside temperatures fell into the teens so likely one more day would have brought disaster.

  40. I could not agree more Matt,

    I am using Google Documents, Gmail and Google Calendar for work and personal life and I never lost any data that way, I do backups of the documents time to time though (in case “the big crash” happen!)

    In the past, I used to lose several days of work because of hardware failure or virus … This issue is gone now thanks to the cloud services

    P.S. a backup feature (save all docs from one account into a zip file) of google documents and even Gmail would be simply a GREAT feature.

  41. Matt, I’d like to buy that dead hard drive from you. ๐Ÿ™‚

  42. This is a topic that I have personally struggled with — especially photos. Should I be self-reliant, possibly risking my precious pictures during a hard drive crash and implement an annual backup process (http://www.somelifeblog.com/2008/01/annual-digital-photo-backup-process.html) or should I rely on a system such as Flickr or Google’s Picasa to do it for me.

    Thus far, I have elected to go with a self-reliant system for two reasons. 1) I know those DVDs I burn will still be in a safe 5 years from now and 2) I have no idea (no offense Matt) if Google or a relatively free storage service that honors my privacy and rights will still be in business or have such an offering.

  43. Hi Matt – Are there plans to allow PDF’s to be stored and searched in GoogleDocs? This would be a great addition and allow users to use GoogleDocs as a virtual filing cabinet and store all their paper docs in the cloud.

  44. Hello Matt,

    I just did a keyword search for “IT cloud services,” and apparently Google still ranks this 2008 blog post very highly — your proven SEO skills at work, no doubt ๐Ÿ™‚

    Anyway, 2009 seems to be the year when IT managed cloud services truly evolves to the next level — both for consumer applications, and within the SMB or enterprise. Perhaps the basic back-up application will be the one that exposes the inherent value proposition to the mainstream user. Then, many more people discover all the other related applications.

    David Deans
    Business Technology Roundtable

  45. Glad you got your data back! It’s not just you that power failures affect. Cloudserve, a leading provider of cloud services in the UK., utililses a number of data centres with real time SAN to SAN replicaton between them to ensure full failover capability. Back in Nov 09 a whole Tata Datacentre in East London went lights out when the power failure, the UPS’s died and the generators failed. Luckily they got the power back up before we needed to invoke a wholescale failover to a second datacentre but juts goes to show you need to be sure you have a backup plan or use a cloud provider that has one and can proce it!!

css.php