It's all got complicated: The costs of data recovery Data protection companies have multiplied in the past 10 years and they are locked in a bitter battle for market share. Before Amazon arrived on the scene, the main options were owning all links in the data protection chain or renting or leasing appliances, with or without attached offsite services. Thanks to Amazon-style …
This post has been deleted by its author
Hear. Hear.
That is a very good description of what we did when we took up the maintenance of the IT foe a couple of small businesses. The main difference was that we used USB hard disks for the archive because only a small number of machines had optical drives and they were CD players (if it aint broke don't fix it).
This post has been deleted by its author
You hit the nail squarely on the head.
I have over just the last 3 years done a number of File System assessments, only to discover that as much as 75 percent of the data, in any given FS, has not been accessed recently. By recently I mean neither accessed or changed in over 180 days, and many for over a year. The customer then has the light bulb moment that they really should look at archiving and some sort of HSM solution. This they discover remove 70 percent of unstructured data from the backup stream and frees up an enormous amount of primary storage as well, thus killing 2 birds with a single stone. It is relatively easy to make the case for archive if they don't have to also purchase additional primary storage every year.
Very, very few places, especially small and medium enterprises, actually need these ponsy overpriced 'cloud' backup services.
Small is precisely where cloud services work best, providing automatable off-site backup where there isn't a convenient existing off-site location to back up to. It's as things get larger it becomes less attractive as the costs and bandwidth requirements go up. Similarly at the high end of the scale there's no substitute for a properly managed archive, but disc but tape - even blurays are a little on the small side and need a lot of cataloguing and management if you are to actually find what you want in a library of thousands of discs. Consumer bluray drives are also completely unsuited to buring dozens of discs a day adding further cost. Archives need monitoring, verification and duplication and those administrative and management costs can easily end up dwarfing costs of anything else if not managed properly which is what you want to avoid if possible - the larger operators can't avoid it, but it doesn't make it the most attractive option at smaller volumes.
It's between those two extremes that things get a bit more problematic and it's where many operations will find themselves - in round terms we're probably talking between 100GB and 100TB. There's no such thing as a one size fits all solution so anything that pretends to be one is going to be inappropriate at least some of the time.
We I was last looking at this around six months ago we determined disk to disk backup on our own servers was the only game in town: we saw that there were two sites just over a kilometre apart with an existing gigabit microwave link between them. Both need fileservers although one uses tens times the data of the other. The servers are needed, the bandwidth was free, so filling them out with extra disks basically meant 25 bay rack enclosures instead of four or eight bay types, additional memory (ZFS likes its memory) and of course the disks themselves. The servers at each end are currently filed with 6x4TB drives giving 16TB of usable storage after RAID6 but with the spare bays and increases in drive capacity they could well be extended to 100TB or more come decommissioning.
We have all the benefits of one big storage array on each site: snapshots are taken literally every ten minutes giving a high degree of resiliance to things like Cryptolocker and the two servers rsync themselves together every couple of hours for site backup. Bit-level verification of data is done weekly. We haven't fully exploited the possibilities for dedupe except for VM images but the opportunities are there. We have fine-grain security control allowing different access controls and encryption dependign on the data. We got all of this for perhaps an additional £800 capital spend and 100W power at each end. Critically, all of this is completely automatic, it does not require an operator to do so much as stick a tape in the drive every day yet alone burn a dozen blurays.
So, yes, that fits us very well but no it wouldn't be for everyone for one reason or another. Your one-size-fits-all approach equally wouldn't suit us - it would be vastly more costly and less convenient for less capability. That brings me back to my first point - universal approaches are nothing of the sort because the underlying requirements vary so much.
"Very, very few places, especially small and medium enterprises, actually need these ponsy overpriced 'cloud' backup services."
Actually, all of my customers tell me the the opposite. Especially the ones who have lost data because they didn't have a proper efficient backup strategy working.
The vast majority of my customers are below 20 employees in density and the company owner doesn't give two hoots how his backup is handled, all he wants is for someone to shout at if there's a problem, and for his business to work for as long as he/she wants it to. Having recently handled a data recovery job for a new customer who was happy that everything was being backed up on his NAS, with mirrored drives, as soon as the main chassis died we then had to swing in to action and save the day. The local IT tech got canned and we won the contract for online backup. Daily reports, managed service (we fix it when they change something which stops the backup from running) and they are a million times more happy.
We only started providing cloud backup a couple of years ago because we realised it was the cheapest way of helping our customers be good at backup. Besides, it's all good and well grumbling from under your bridge about how easy it is to backup and small business don't need it, they just need to... ah, there's the prerequisite. And if the companies don't meet your defined prerequisite of actually being good with organising backup, then they DO need someone else to do it for them.
I agree generally with the arguments at the head of this thread. The problem is that the real world gets in the way, especially with small outfits.
The good IT bloke or blokess leaves, and another lesser mortal is hired some time later. And with him/her, knowledge of the backup regime withers.
Or what the management consider to be a good IT person is not actually a good IT person. And how would they know? The IT bod claims there's a back-up strategy, but nobody thinks it apt to regularly test it in anger (which is the only way to know if your back-up strategy is a recovery strategy too!). Then, the shit hits the fan and management lose faith in DIY techydom. Not that cloud is better. You still need someone competent to use cloud services to keep your company's data safe.
I know a company whose head had no idea that a major project's assets were stored on a single spinny-disc with no back-up strategy. Luckily, the project had failed for other reasons, because when the spinny-disc finally failed, the head of this little business asked me if I could get the project off the backups. What he meant was 'off the tape thingy in the rack' - a back-up system that had packed up ages ago - and even when working was storing an ever-more truncated subset of said project
The thing that failed wasn't the machinery, but the people supposedly in charge. IT people came and went... 3 or 4 times. The last person whose job it was to run things wasn't even aware of a backup strategy, and there was nobody in the building who considered this an issue.
So no. Cloud isn't the answer. Neither is on-site backups. The answer is a management who know to demand technical competence, and require their staff to demonstrate said capacity. If you don't know, you won't ask, which is why so much is screwed.
I would agree with you but for one thing : if management has no clue, then the "solution" employed doesn't matter ; whatever it is, it'll get screwed up.
So forget the clueless management, they're doomed anyway. We're talking about backup solutions that work, which implies management know how it works and, more importantly, why.
And for those cases, cloud simply does not cut it.
However, I fully expect future disasters to be laid squarely at the feet of incompetent management, so saving the bacon of cloud providers despite the fact that they cannot possibly deliver on the bandwidth and availability they promise because they do not control what happens between their cloudy servers and the client's door.
This post has been deleted by its author
Exactly - you can try to make a case for 'cloud' backup out of crap management, an ever changing line of incapable IT managers or any number of other excuses. But that won't change the fact that you're outsourcing a critical part of your infrastructure, and sooner or later it will come back to bite you.
Compare that with your following comments:
On the other hand, I have NEVER failed to provide a client with ONE BIT of their data when I have been responsible for their backup routine. Never.
So, if people outsource their backup provision to you they are being very responsible. If they outsource it to anyone else then they are being reckless? That doesn't appear to be a position with any great principle to it. What makes your own methodology so solid? Oh, apart from your own incompetence:
AND I've gone beyond my obligations too, when a tape wouldn't read, I didn't fob them off with "last week's copy", I retentioned the tape over and over, and cleaned the heads, I've even had old QIC-02 drives in bits before now, and gone as far as reading other client's tapes in my personal machines, anything to get their data back.
Good for you. Your backup methodology was flawed but you managed to recover the situation. That is good luck rather than good judgement. This would never have happened in any place I've ever been using tape: if Friday's full backup doesn't work no matter, you go back to Thursdays. Then you apply the incremental you took at close of play on Friday to recover to where you should have been. No amount of retensioning or swapping the tape between drives will recover the situation where an eighth of an inch patch of oxide has simply flaked off the tape never to be seen again. And yes, I have had that happen. You backup systems aren't bulletproof, you have simply been lucky.
Storage isn't easy for some workers even if the knowledge of the back up regime is meticulously laid out in check sheets and detailed procedures. There are people who can do the thankless, invisible, yet often critical jobs and they will be seen by good management as valuable workers. Thing is most do not have good management.
A person spending their time on creating, developing, using and tuning a good back up regime often thinks they are going a good job and feels the hero when an otherwise impossible request is filled almost instantly.
But in a company with poor management they are not doing a good job. The almost instant retrieval or recovery will not be noticed, or worst yet valued, and the time spent on the back up regime is time not spent assuring management that everything is being done to the maximum possible. The time spent with management will also reap rewards when things go wrong and an excuse they will accept is needed.
Short answer? Because if you are only backing it up to a box on your own premises you aren't disaster proof. (Unless you're using an IOsafe as the target.) Backups are complicated, mmmkay?
Backups are indeed complicated, and the first thing to do is work out what your needs are. Do you even need to be 'disaster proof" ? Start with the analysis, which data are critical, how much can you afford to lose before the business is impacted, how long can you afford to be without them, etc. The results of that will tell you if you need dedicated live synchronous replication across multiple sites, cross-your-fingers cloud, or just a DVD in the safe at home once a month. Part of that analysis, of course, will involve looking at the Data Protection requrements for the info you need to save, where you can legally put it, and how concerned your customers might be if they don't know exactly where it's being kept.
Starting with the solution ("put it in the cloud") and trying to make the problem fit that solution is a recipe for disaster, not one for disaster recovery.
When have I ever started with the solution "put it in the cloud"? Hmm? Do you even read the articles anymore? Or you just see "cloud" then whip your your Mighty Ragekeyboard of Justice? Dear $deity man, I have a pretty solid track record as being one of the most anti-cloud writers The Register has! I have written any number of articles warning about the many and varied dangers, and not to take the supposed cost savings at face value.
Despite that, the cloud is worth consideration. It is - for backups - probably even cheaper (under the right circumstances). It's one tool amongst many, but it is an increasingly valid and important one. Even to "cloud haters" like me.
What shocks me is that you are so averse to cloud computing that you would advise customers to be prepared to simply lose data rather than use the cloud. And, of course, you're assuming that the customer has sorted and organized all their data or has the time to do so. In many cases - in my experience, in most cases - it is actually cheaper to just buy the extra damned storage space than it is to throw umpteen man-hours at an organization and rationalization project.
While I agree that not everything in every company has to be backed up, or made disaster proof, not every company has the means to make that determination, either. And - to be perfectly blunt about this - not only are you better safe than sorry, but the cost of cloud storage just isn't high enough to piss and moan about any more. The cost of data loss, however can be the whole company.
So, while you and I could sit here listing exceptions to every rule, the existence of an exception does not invalidate the rule. The rule is: everyone needs backups and everyone should be sending some or all of those backups to a disaster recovery site. The addendum to that rule is that cloud backups are generally cheaper and certainly way the heck easier for most companies, like it or not.
Now i personally don't like it - I have massive data sovereignty concerns with cloud computing - but my personal preference doesn't change reality. And I submit to you sir that your preferences don't change reality either.
The elephant in the room with the cloud that no one wants to look at is the physical access to said cloud. It might be fine in the cities where you have a high speed link but there are places in the country where that will never happen.
We have a client that was 'persuaded' to use the cloud for storage and backups. The problem was his ISP could only give 512kbs so backup took over 10 hours if they didn't need to use the connection for anything else.
When we took over looking after the IT there we rationalised just what they needed to backup and what they needed to store. A second-hand storage array and a server obtained from a large company that was upgrading brought what they needed in house. The addition of a wireless link gave them off site backup. We do remote checking of the logs overnight and if there is any question or apparent problem I send someone there in the morning.
Result, one very happy client that is saving money compared to what the consultants insisted he needed.
As I said at the start, the cloud option may well work and be cost effective for those that have a high speed link but not everyone does and it is those that don't that also need consideration.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
