In this podcast, W. Curtis Preston, backup system expert and executive editor for TechTarget's Storage Media Group, explains the basic goals of a sound backup strategy that channel solution providers can apply to their clients. Read the transcript below or download the podcast.
You must have Adobe Flash Player 7 or above to view this content.See https://www.adobe.com/products/flashplayer to download now.
Download for later:
Modern backup tools and technologies
• Internet Explorer: Right Click > Save Target As
• Firefox: Right Click > Save Link As
What technologies should an integrator or VAR be familiar with when discussing a modern backup system with their customers?
It's definitely moved past just [the need to know] a few backup packages and being familiar with modern-day tape drives. They should be conversant in CDP, deduplication -- both source and target deduplication -- and also near-CDP and replication. All of these can be part of a modern-day backup system.
Why is disk such an integral part of a modern backup system?
The problem is tape. Tape is great. It's cheap, it stores a lot of data, relatively reliably -- I would say less reliably than disk, but in terms of the gigabytes and gigabytes and, really, petabytes that we've stored on tape over the years, it does a good job. The problem is that tape has gotten faster and faster, and it's a problem of the industry that tape drives had to get bigger to sell more tape drives, so just because of physics, as tape drives get bigger, the tape drives get faster, and as they get faster, they get harder to keep happy. Well, disk drives, unlike tape drives, are happy with fast and slow. So disk drives can be used as a buffer to tape; that's their initial primary use by a lot of people. And then secondly, the thing with disk is that there are a number of technologies, such as CDP, near-CDP and replication, that are only possible if disk is the target.
How do CDP and near-CDP play into the modern backup system strategy?
It's a little difficult to answer in this format given the time we have. But I would say that it should be something that's examined. CDP and near-CDP allow for things that are simply not possible with regular backup. You can do backups throughout the day. You don't have this concept of a full backup every once in awhile, or even a nightly backup. It's a backup that's continuously happening throughout the day. And the differences between the two: With CDP, you can recover to any point in time; with near-CDP, you can recover to significant points in time, such as an hourly snapshot that you take. The nice thing about both of these products is that they can offer some really interesting backup possibilities to some companies; if they're smaller, they could really consider using this pretty much for everything, especially if they tend to be Windows-centric environments because one of the challenges with CDP and near-CDP, with some of the products, is that they tend to be Windows-centric, and they forget the other operating systems and applications that are out there. Larger companies use CDP and near-CDP to meet specific requirements that they simply can't meet -- for instance, a database that's simply too large or an RPO, recovery point objective, that's simply impossible to meet with regular tape or disk backup.
What's the drawback of CDP and near-CDP? Are they expensive, difficult to implement?
Well, it's mainly [an issue of] familiarity. So people know backup; they know tape. They get it; they know how it works. CDP, for all its wonder, is weird. It's just completely different from what people are used to. And it's harder for them to grok it, to get that concept, and they don't like things that they can't understand. It seems new and risky. And it is new and risky. Comparatively speaking, backup has been around for decades, and CDP [and near-CDP have] been around for a couple of years. So really the biggest challenge is simply getting the customer comfortable with the solution. As to whether it's more difficult to implement than regular backup, an argument can be made both ways. It may be more difficult to implement, depending on a number of factors that we don't have time to go into, but it can absolutely be actually easier to implement than [a regular backup system] -- and certainly the restores can be much easier.
What about deduplication?
Dedupe is a real game changer. If the customer is a more traditional customer and they want to stick with the traditional backup products, dedupe is a way to use disk as a primary target for significantly less money than it would cost otherwise, so it's a real game changer. And it also allows a person to continue to use traditional backup if that's what they want to do, and still replicate that backup offsite, which is just a wonderful thing. If they want to go a little farther down these option curves, they also could consider source dedupe, which is backup software that has dedupe built into it. You could use that to back up remote sites, mobile data -- basically data that previously had either been ignored or had been backed up in ways that we don't talk about because it makes us feel uncomfortable. So dedupe can make backing up that remote data and mobile data much, much easier and much more secure than ever before.
And much cheaper too, right?
Yes, much cheaper, I would say if they were doing [remote site backup] right before, absolutely, dedupe would be much cheaper. CDP and near-CDP would also be another way, because they are also incremental forever and can also be used to replicate but both of these technologies, if they were doing the right thing before (which included local tape-based backup and daily trips to an off-site storage vendor) then absolutely, this is going to be less expensive. It may or may not be less expensive if what they were doing before was not the right thing to do, such as a local DAT drive that would just get repeatedly overwritten every night. And you do find that in many companies; the remote sites tend to get ignored, and their backups tend to really stink as a result. So it is cheaper than doing the right thing without it, but it may or may not be cheaper than what they're doing right now.
So do you have people who come to you who are disappointed when they figure out what the reality is in terms of cost savings?
Oh, absolutely. Backup is one of those things where it's the back-of-the-bus kind of thing. They'll spend all kinds of money to make their application faster, or all kinds of money to do whatever, but [a backup system] has to really prove itself, and in the end, all it really is is an insurance policy. But I have very little patience with customers that want to justify to me, "Well, we'd rather continue doing this thing" that has all of these horrible, horrible things attached to it. No one could possibly defend continuing to do it this way, but it does have the one feature that it is less expensive. But the only thing you can do at that point is sell risk and the concept of risk. Do you really want to be that company that loses data, [loses control of it and the story ends up] on CNN. If you're doing the right thing, your tapes are being moved all around the planet if you've got remote sites. If you're doing the wrong thing, you could lose the data -- say, if you have a disk drive problem and you've lost the data -- depending on what type of company you are, that could either cost you significant amounts of money or [it could] also be a public event. Do you really want to be that company?
How do designs for data centers and remote offices differ?
In the data center, you have typically all the bandwidth you want. You have Gigabit Ethernet, we're starting to get 10 Gigabit Ethernet. If you don't have enough bandwidth, it's simply a matter of buying a new switch or multiple switches. [At] the remote site, bandwidth comes at a cost. I've had clients at telecom companies, and they tell me that bandwidth costs money to them as well. So you pay for bandwidth at the remote sites as a service. The bandwidth is nowhere near what you have available in the data center, so you've got to take that bandwidth into consideration. So if you're considering any sort of remote backup ability, you've got to have a [backup system] that takes that small amount of bandwidth into the equation.
What does that mean in practical terms?
It means that you have to examine CDP, near-CDP or dedupe as your way of doing remote site backup. It is no longer necessary or considered practical to use tape [for] the remote site. I haven't checked the stats lately but it used to be DAT drives, DDS drives were the No. 1 installed tape drives in the world, mainly because of remote sites. … And that used to the way you did it, and the customer had to figure out how to move those tapes around and secure those tapes. That is no longer acceptable for a number of reasons. Tape drives need to be encrypted if that data's being moved around. More importantly, there are options now that allow you to back up that remote data, both locally to that remote site for faster recoveries, and also remotely to a central site to secure in case of disaster recovery. And those solutions include CDP, near-CDP and deduplication.
And so you wouldn't have a situation where you have a random employee at the remote site responsible for handling tapes?
No. One of my mantras with regard to backup [systems] is, the more you can get the human out of the equation, the better. And that goes for trained IT personnel, and it goes double for the non-trained, non-IT person at the remote site who can't even be bothered to swap the tape drives out. I've got many, many stories of what happens when you put that responsibility into the hands of people who see it as an annoyance.
Should reporting and monitoring come from a backup vendor, or should it come from another kind of vendor?
Well, I think that if a customer is a customer of a particular backup product and that company provides a reporting product, they should examine that one first because it is possible that they will offer more advanced, increasingly integrated reporting than anyone else because they know that product. But that doesn't necessarily mean that they will have the best reporting. There is an entire industry of companies that do nothing but monitoring and reporting. And I would recommend that they also look at the chief competitors in that space and examine how those products work. I will say you get what you pay for. Some of the backup software products have reporting built into them and it's free. And some of the same products will offer an advanced reporting tool that you pay extra money for. If you're going to be paying extra money for it, you should also examine the competitors' products and see what they offer.
Would the built-in reporting and monitoring tools ever suffice for a small company?
It's a challenge to say where the line [is to determine] when a reporting tool is sort of mandatory. The real challenge for a small company is that even if they had a reporting tool, who would be looking at it? So generally speaking, it depends on what you mean; is it a small company with IT staff, or is it just the guy who happens to do computer stuff? … For what I call a TSB, or truly small business, they should probably outsource their backup services altogether, to a company that just does backup management. That way that company cares about them and their data, and they will notify them when there's a problem. If they're a small to medium-sized business, they start to get to the point where they should consider doing their own backups and doing their own reporting. And really the question becomes, Are they getting proactive notification? Some of the things that happen with larger reporting packages include things like, when the backup person isn't doing their job, for example, if the person isn't monitoring things and thresholds happen, like more than 20% of the backups are failing, or the server goes three days without a backup, or something that should cause somebody to be upset, reporting tools can bubble that information up. It can escalate it automatically and send an email to the boss, and the boss can say, "Hey, what's going on over there?" The biggest problem with backups in both small and large companies is invisibility. So [a reporting tool] is a really easy argument to make when you're a large company because you tend to have multiple backup servers; you also tend to have multiple backup products. So it's a really argument to make for a reporting tool at a larger company. At a smaller company, it's going to have to be a call that they're going to have to make. But frankly, I'm a fan of completely outsourcing that entire [chore] to a service.
Go to Part 2 of the Channel Expert Podcast, on backup testing and verification.