There are three types of scaling that matter to today’s sysadmins, however, only two ever get talked about regularly. When talking about hyperconvergence, everyone is familiar with the idea of scaling out; adding nodes on an as-needed basis is pretty much the whole selling point of hyperconverged infrastructure.
Some vendors – and SimpliVity gets a shout here – allow for scaling up. Scaling up is the ability to upgrade nodes by adding CPU, RAM, networking, storage or all of the above. This can come in the form of field-upgrades to existing clusters, or merely flexibility in the purchasing process allowing for variability in meeting customer needs.
But there is a third, very significant, type of scaling: scaling wide. The ability to scale wide (also sometimes known as geoscaling) is the ability to scale a solution across multiple sites.
Not to put too fine a point on it, scaling wide is a pain in the proverbial. No matter how you want to approach the problem there are always limitations. The software is usually not up to snuff. The WAN links are never big enough, and there’s that pesky speed-of-light problem that we have yet to overcome.
Searching for real world solutions
What we want is a magical solution that does real-time replication between sites. I make a write to a VM in New York and that write also occurs in Tokyo. Recovery Point Objective (RPO) of zero seconds. The holy grail.
Sadly, in the real world, this is spectacularly expensive. Even when money isn’t an object, when you can buy the biggest WAN pipes in the world, afford the most expensive tech from the vendors who overcharge just because they can…it still doesn’t always work.
Traditional vendors don’t have a lot of options between “give up on the multi-site idea” and “if you have to ask, you can’t afford it.” What they do have is usually some evolution of a traditional backup solution. These are expensive, and evolutions of ancient code bases with layer after layer of filtering and feature cruft grown upon them.
These backup-based solutions claim to be able to back up your physical and virtual machines to some other site and then light them up on a virtual infrastructure. Some times it works, sometimes it doesn’t. When it does work, it’s usually because an enormous amount of effort has been put in by both the vendor and the IT teams to make it work.
These backup-based systems are also incredibly fragile. Look at them wrong and something goes wrong. These impose sever design limitations, change management concerns and more.
This might be fine if you’re one of those unicorn companies where money doesn’t matter, with a stack of idle sysadmins. Unfortunately for us sysadmins here in the real world, we don’t have those kind of resources. Our hair is usually on fire, and once we set something up we probably don’t poke it until it breaks. We’re too busing fixing everything else that’s broken.
This is where SimpliVity comes in. Remote office branch office (ROBO) infrastructure is what they do. No, it isn’t magical “0” RPO supersauce. But neither is it the “trying to create the Sistine Chapel in crayon” that backup solutions offer. SimpliVity has site-to-site replication baked into their product and they’re probably the best currently available solution for the job.
A lot of the hurrah for their solution comes from their data efficiency technology. They have a hardware accelerator and some neat intellectual property that means they do data efficiency really well, without impacting the performance of workloads.
The knock-on benefit of this is that SimpliVity is super efficient at moving things from one site to another. This is because they only send what absolutely needs to be sent, dramatically lowering WAN usage, and allowing more sites to be connected using the same bandwidth.
SimpliVity has also made scaling wide their focus. They’ve doubled the number of sites that they can support and included dynamic network discovery, including a hub-and-spoke topology that further enhances distribution of data throughout the network using their data efficiency technology.
SimpliVity provide a solution with the lowest RPO that is reasonable to expect for most businesses. With SimpliVity, businesses can light up VMs from any site on any other site. It’s the same technology on both ends, so there’s no guesswork around disaster preparedness.
When it comes to solving very real pain points – site to site replication and migration – SimpliVity offer a solution that’s more reliable than the competition, generally less expensive and hands down the easiest to use. That’s a discussion point worth dwelling on.
Yes, what SimpliVity does regarding scaling wide is “just a feature.” It’s also just a feature that nobody else can touch them on. As everyone else scrambles to pack all the features into their solutions, it’s worth celebrating those solutions which exemplify the best of those features. They are the barometer by which all other solutions will be measured.
About Trevor Pott:
- Explore more of Trevor’s thoughts in his personal blog.
- Read Trevor’s blog on scale in The Register.
- Tweet your thoughts on scale to Trevor or SimpliVity.