Over the past handful of weeks, I've written about the VSX and how it's able to work its magic. Today I want to talk about two other tricks: how it provides no single point of failure, so your data is always up, and how it saves you time by using snapshots.
No Single Point of Failure
The VSX accomplishes this with two elements: high availability and mirroring. While a data center can opt for a single VSX, most folks get them in pairs. Two VSXs, a primary and a secondary, can talk to each other. The secondary is continually watching the primary to make sure it’s okay. If it senses there is something wrong, the secondary shuts down the primary and takes over the work.
The switchover happens with almost no disruption to the network. There is a small delay as the initiators figure out which port they need to be talking to, but it’s not noticeable.
So that’s the first way in which the VSX provides no single point of failure—high availability through the redundant hardware pairs.
The other way the VSX creates a no single point of failure system is by backing up to your local machine by mirroring the physical volumes behind the scenes. I’ve mentioned in previous blog posts that underneath a virtual block is a physical block. The VSX can mirror each block on a separate SRX, which essentially creates two SRXs that are copies of one another.
In that case, if you lose a controller, for example, the VSX still can get your data. When you recover the controller, just retrieve the backup along with the changes that have happened since it went down.
Redundant hardware pairs and mirroring in the back end means that if you diagramed a few SRXs and VSXs and throw in a couple of Ethernet Switches, you can’t put your finger on one element that could give up the magic smoke and bring your system down.
If a switch goes, well, there is another one. A VSX fails, there’s another one of those too. Mirrors ensure that if an SRX dies, the system is safe. There you go. No single point of failure.
A local mirror is perfect for local backup, unlike RAID, but for disaster recovery, many organizations shadow off-site, which means you are protected with a remote backup too.
Let’s imagine you have a lot of, say, Windows operating system. To save time, you want to create a golden image that is almost completely configured with all the stuff you have to do for every instance in your shop.
With the VSX, create a virtual disk and install Windows on it. Once you have it set up so it’s the way you want it if someone were to request a Windows server, make the disk read-only so no one can modify the volume. Now, if someone needs those bits, use snapshot to create a read-write clone.
Just like with other snapshots, you have one copy of the data but two volumes pointed to it.
When those bits are needed, install the snapshot and make all the local configurations you need. Hank’s computer can have the weird thing he asked for, Suzy’s gets the stuff she needs, and all of accounting has their special tools.
There is only a single copy of the operating system you pulled off a cd. Instead of 47 copies of Windows, you have one. Everyone has their own clone of it. By having 90% of your configurations done before you even begin, snapshots make your life easier.