Press Contact:
Stephen Lawton
Acronis, Inc.
+1 (510) 285 1055, ext. 7030
stephen.lawton@acronis.com

Original article on Let's Talk Computers website

Technology Trends in Virtualization

Complete Transcript of Acronis Interview Ed Harnish — Acronis Host — Alan Ashendorf on Let's Talk Computers June 6, 2007

Alan: One of the new "buzz words" within IT is virtualization. Why are more corporations, turning to this exciting new technology? Joining us from the Microsoft TechEd Conference in Orlando is Ed Harnish, Vice President of Marketing with Acronis. And welcome to Let's Talk Computers, Ed.

Ed: Well, thank you Alan.

Alan: Ed, first of all, what is TechEd and why is that so important to Microsoft and to IT?

Ed: TechEd is one of the premier conferences that Microsoft puts on. This year, they are doing three historical ones in North America; they've got one in Barcelona, and I believe that they're doing one in Singapore. It's a great place to find out what's going on, new with Microsoft, see the new technology, (to get your eye-balls on some new things). It's also a great place for training and testing.

Originally it started out was for SE to become MVP or MCSE and to get their certification at a single location. It really has a technical bent; a lot of IT people are there, technical people who are looking for new technology that they are testing.

Alan: One of the major tracks that Microsoft has at TechEd is virtualization. Why is that so important?

Ed: There are a number of reasons — if we are looking at the larger companies, the Fortune 500, the Fortune 1000, we are dealing with the data center consolidation issues, whether that's cooling, they simply can't cool those rooms with the processors running so hot; cost of electricity. Another thing that's becoming popular today is reducing the carbon footprint, as people say.

But, they're really trying to reduce the physical number of servers they have in a data center. They're looking to do that cost cutting, as well as for the ecology. One other reason that they would do it is that it is easier to manage. It could have more servers in a single, physical device so I can go to that physical device and manage those four units, rather than having to log into and actually attach to four separate servers.

Alan: I know one of the hardest jobs that an Administrator has is taking care of hardware. And hardware can always trip you up. This is where virtualization really makes it possible to have one major piece of equipment and then you can run any kind processes on top of that and it's more efficient, isn't it?

Ed: It's more efficient at that single location. A lot of people will say, however, that you have a single-point of failure, now. Right, where you had four servers — if one server fails in a physical world, then the other three would operate. If they consolidate those into a single, physical device — if one server fails; if a power supply goes in that server, they all fail. That's the down side.

The plus side is that by design in nature, virtualization and virtual machines are very easy to move, copy, and get up and running so that people can very quickly, much faster than in the physical world, bring up another virtual machine, if in fact, that power supply fails on that first device. So, your consolidating really out-weights that single-point of failure issue.

Alan: This is where back up and recovery becomes critical, because with multiple processes running on the same box, you cannot have any down-time, whatsoever, can you?

Ed: That's exactly right, Alan. So, if that power supply fails and now, all of a sudden my Accounting System, my Web Server, my Exchange Server and my General Purpose Server fail because they are all under that system, the entire company goes down. So, back up and recovery is very important for this process.

Alan: Well, you are doing a presentation in TechEd on Virtualization Gotcha's. What are some of the Virtualization Gotcha's that IT needs to be aware of?

Ed: Some of the things that we have found that they need to be aware of are things like the amount of resources that are used; and how resources are used. If we think about in the physical world or what is on a physical server, a single server was set up for a CPU to run one operating system — with one set of applications with one set of users.

Now, we're asking that same CPU to handle multiple operating systems, multiple sets of applications and multiple sets of users. One of the Gotcha's is a lot of people look and they say, "Gee, I have this old server in the corner that we haven't used in a while; I'll make that virtual. One that happens with virtuals is — with more memory, faster processors, higher performance disk — the better off you are. So, one of the things people do is underestimate the amount of processing power they need.

As an example, if I were to set up a virtual PC, let's say, running NT 4.0, as a guest, you need about a Gig and a half of disk space, 128 MEG of RAM in order to get that running. You need an additional Gigabyte of disk space and 64 Meg of RAM to get that first VM (virtual machine) running. So, we're talking about 2 Gigabytes and almost 200 Meg's of RAM, just to turn that system on. So, one of the Gotcha's is that they do underestimate the amount of space, the amount of processing power that's required.

Alan: How is Acronis helping us to be prepared for this new world of virtualization?

Ed: We are helping in fact is by with our backup device, we're in the virtual world, of the virtual machines. It's much easier to move it from one physical machine to another. In a physical world, if I took an image with that of the operating system, applications and all of that information off a single server, most of the time we have to put it back on that same server or go through a lot of maturations to change drivers and identify the differences between those two machines.

Since, by nature, a virtual machine is insulated from the hardware, via the virtual operating system sitting on top of it, you can pretty much take a virtual machine with a physical server and move it from one device to another, without having to worry about the underlying hardware.

What we do to facilitate that movement, so we're able to capture an entire virtual machine or segments of that and stream that to another server and get that up and running. What some people will do is to take a server, create a virtual host operating system on top of it and set it in a corner, (kind of a hot stand-by), if you will. Then, if any server fails in that system, all they need to do is take a back up with Acronis True Image; they would have one anyway, right? They'd be taking backups as they had done and restore it to that stand-by machine that they have.

Alan: Well, this is where IT's eyes roll over in their head, when you're talking about restoring a virtual machine to a physical machine or take a physical machine and put it into a virtual world. You have a lot of Gotcha's that you really need to be aware of and organization is very important at that time, isn't it?

Ed: Absolutely, Alan! You, know, a big one of those is that today, the current state of processing power, things like Exchange, SQL, Oracle — anything that's very I/O intensive or makes kernel level calls, doesn't necessarily virtualize well. The picture that you had a forth VM, under a host operating system and one of them was running SQL server. And you had a multi-million record database that you went to index. Well, that's a very CPU-intensive task. So, while that is re-indexing that set first machine, that SQL server, you would see degradation across those other three systems.

So, a performance tuning capacity planning is a big Gotcha. And unfortunately, today there are no predictive tools or capacity planning tools, in the virtual world that will say — okay, if you are running ten physical servers with these five applications, you can put that into a single VM. It's really a trial and error basis if that happens.

So, often times that IT staff absolutely says, "We're going to have to move that database, whether it's Oracle or SQL server back to a physical world, because my users were complaining that once we went virtual that they saw degradation, across-the-board."

Alan: One of the big problems about moving any kind of computer or backup to another computer system that has dissimilar hardware, a lot of times it won't boot, will it?

Ed: Absolutely not — in fact, you can go to a vendor and buy two servers from a vendor today; have them both shipped the exact same day and there are enough differences inside that machine that if you took a backup from one and tried to restore it to that second machine, it simply wouldn't work.

The people that are looking at a machine that are one — two — or three years, the idea of calling up a Dell or an HP and saying, "I need that exact same machine" — it's impossible! It's simply not going to happen.

Alan: A lot of times, the computer system may be completely wiped out and you have to start over from scratch and getting back up, that is the most costly part of IT or for any kind of corporation — a computer system goes down; they need to get it up and running as fast as possible. And how does Acronis help us do that?

Ed: Alan, what we say is, "It's all about the recovery". The backup — we've got it figured out. It's schedulable — I can say every night to do an incremental and on every Friday night, do a full backup. Every Thursday, backup my mail system and the system wakes up at Midnight and does it. It's schedulable, process tolerant; I can handle that; there are good interfaces for that. The crisis comes Friday afternoon, the last day of the month — 4:00 p.m. my accounting system fails. Every CXO in the company is calling, saying, "I need that system running. I need to get it up and running".

So, what Acronis does is really to concentrate all of its engineering effort on that recovery process. So we allow things like recovery in background — so that while the system is coming up and running, people can get on it. In fact, we go one step further — we allow you to prioritize the recovery. So, that example that I gave you — It's Friday afternoon and the accounting system fails. The IT staff would then start the restore process and the applications' names would be mounted; they would see the name; they would pick the accounting files, (the accounting program, Solomon or Great Plains, whatever they are using) — click on that.

We would then restore the accounting system first and all the associated files and data; let the accounting group get working and then in the background, restore the rest. So, if it takes two hours to do a complete restoration of a server that's completely gone and fifteen minutes to get my accounting system up, we allow you to prioritize that application; bring up accounting first — now they're happy and then I continue to restore the system in the background.

Alan: Well, Ed, that sounds very complicated. How can an IT manager set that so it happens in the background?

Ed: In fact, it's transparent to an IT manager. All they would have to do is when they start the restore, it will actually ask them and prompt them "Do you want to prioritize restoration?" If you say, "Yes" — The next thing it will do is it will to show you a list of all the available applications. You literally put a check box in the one you want and say, "Commit". And at that point it does it all in the background; it will know to pull that first one back and pull all the rest under the covers.

Alan: Well, as you pointed out, restoring is proof of the pudding, because if we can't restore, we have nothing. How does Acronis help us to make sure that all of our backups have been totally verified so we know we have absolutely, good backups all the time?

Ed: That is something that worries all IT staff. The first thing that you want to do and that we recommend is a set of back step practices. You set up a grand father, father, son rule, where you actually have three sets of backups and all of those are really generational, if you will. So, if one fails, I can go to that second one or go to that third one.

After that happens, we are doing a verification; after that backup and in fact making sure that what was on that server is not only the same thing that's on that backup device, which is very important — but it's also readable. Just to say that it's on that tape doesn't work if, in fact, when I try to restore, that tape is unreadable. So, our verification will first off say, or checks to make sure that everything that you wanted to back up, got backed up and then secondly, it will allow you to recover a piece of it to make sure that that information is actually readable.

Alan: There's nothing worse than trying to restore a computer system and then finding out that your backup is corrupted, halfway into the restoring process. Here, you actually can verify the complete backup, to make sure it's not corrupt, before you actually restore it.

Ed: There's nothing worse. Remember, that this is a crisis point. Everything's down; my company's at a halt; there's no one who's productive and you've got to get that system up. So, you're half way through it and realize that you've got a bad spot or something's not being restored. It can be horrific. So, we actually take that additional step to make sure that it's not only complete backups — that it's verified and readable.

Alan: And where can people find more information about virtualization and also other Acronis Products?

Ed: We have some great White Papers on our site. In fact, we're just finishing up a White Paper that we're doing with Microsoft and Dell, which talks about the migration and maintenance of virtual servers. And we hope to have that out within the next 30 days or so. They can come up to our www.acronis.com to look for that and there is also information about our specific solutions on our site.

Alan: Ed, I want to thank you for taking time off from TechEd to talk to us. We hope to have you back again on the air again, real soon.

Ed: Well, thanks very much. I enjoyed myself, Alan.