CFD3 Prep Post: Riverbed

This is part of my series of posts on Cloud Field Day 3.

I first encountered Riverbed over ten years ago while working at a large Australian telco. We were looking at Riverbed WAN accelerators at the time, and the Riverbeds came highly recommended over the equivalent Cisco product.

From memory, the product back then essentially stripped back really chatty protocols like SMB and NTFS so you weren’t sending a bazillion packets over the WAN links, and it did some fairly rudimentary (by today’s standards) fingerprint-based data caching. The principles don’t seem to have changed much.

Given the amount of data we were trying to send over relatively small pipes (again, by today’s standards), optimising the traffic was something we were quite keen on.

The Story So Far

Riverbed has since grown from that fairly singular product focus to a broader range of products from the looks of things. The WAN optimiser product is now the Steelhead line, which still seems to be at the core of what Riverbed does. It’s added a central management thing called SteelCentral which makes sense when you have enterprises buying lots of accelerators that they then have to manage. Enterprises like to buy management tools to help them deal with scale, rather than building their own things internally, though they do that too. Enterprises also love duplication of effort.

Lots of these point-to-point acceleration points that plug into the network leads to having something that can do more network things, so we see Riverbed has added SD-WAN things of various flavours. That helps Riverbed get into doing more of the core network things that are harder for an organisation to get rid of. If you’re just a WAN acceleration appliance, I can swap you out for a competing product pretty easily, but if you burrow your way into all my other network stuff, it’s much harder for me to extract you from my network without damaging things.

Riverbed also bought wifi network provider Xirrus last year, so that’s expanding out the general network side of things, and wifi is an area with loads of activity these days, almost adjacent to the core networking world that still appears to be dominated by Cisco, though its grip is slipping. The software-on-x86 servers part of networking is slowly encroaching on the way enterprises do their networking as people start to relearn how to use a general purpose computer again. I’m still quite surprised at how long it’s taking, but I haven’t been deeply in touch with the networking world for quite some time so a lot passes me by.

Cloudy Stuff

Everything has to have some cloud in it these days—because reasons—so Riverbed have added cloud-related things.

Mostly this seems to be network- and/or application-performance monitoring things. This is where things start to get a bit… fuzzy… for me, because exactly what Riverbed does with cloud stuff isn’t overly clear to me. I think it’s an extension of the SD-WAN idea: use us, and get a unified network fabric across everything because software. But what that means, in terms of what you actually do with the products isn’t immediately clear.

I’m hopeful that if I did some more digging into the details on the website I’d get more of an idea about what the cloud aspects of Riverbed are beyond “it’s totally a different thing now because it’s software”.

Which leads to something that I will start but not finish: the cloud is ‘just’ the new datacentre. By which I mean if you’re an application developer, or the applications part of IT in an enterprise, the cloud is just one of the places you host applications today. Before cloud, it was in a datacentre run by IT, or (more likely) by whichever large managed services company IT had outsourced it to. The ‘edge’ was branch offices. And still is. Your application stack might have changed a bit—swapping Websphere for cloud-native bits and pieces perhaps—but in a lot of cases all we’ve done is shift the infrastructure piece across so I don’t have to wait four months for IT to spin up nine new virtual machines for my QA environment. Oracle is essentially betting a large amount of its cloud aspirations on this approach.

Which means I care a lot more about how the software parts of Riverbed work than I do about the specific appliances they have. I do care a bit about the deployment options—SaaS vs. PaaS vs. cloud VM vs. self-hosted VM vs. appliance vs. Raspberry Pi—but beyond that it probably doesn’t matter. Except when it does, which should be truly exceptional and I will want to know about it but I will audibly roll my eyes if it’s not actually exceptional and marketing have just slapped a buzzword on something totally ordinary. There will likely be accompanying snark.

Be Prepared

In a show of preparation that few companies demonstrate, Riverbed has done its own prep post. Honestly, I wish more tech companies were this organised in their PR and AR efforts for big conferences. I’m lucky to get an agenda the day I get on a plane a lot of the time.

But, I deliberately didn’t read it until just now so I could form my own impressions first.

The Riverbed post goes into the agenda of what they’re going to talk about on the day, which does follow along with the impressions I had above, happily. We’ll be talking about Riverbed’s attempt to position itself as “The Digital Performance Company”—whatever that is. Zeros that can jump the highest?—and then getting into how the product portfolio relates to cloud. Visibility and workload migrations, apparently. Then we have stuff about end-user experience in cloud apps, wifi and SteelCentral, and connecting to the cloud with SD-WAN. Oh, and we’ll look at Cloud Edge as well.

Pro-tip: no we won’t.

Riverbed have a mere two hours and there is no way we will cover all this stuff in that time. If Riverbed try, it will be a disaster.

I hope someone in the organising team reads this and explains to people how Tech Field Day is about deep-dives into the tech, and if they try to run to their own agenda and insist on clicking through their lovingly prepared slide decks they will get eaten alive if they’re lucky. If they do really badly, the silence will be deafening.

We’ll have to see what the group is most interested in at the time, and Riverbed will need to be ready to jettison all plans once that topic reveals itself. By providing this laundry list of ideas, it’ll probably be something in the first 45 minutes, which means either the application performance monitoring stuff or cloud migrations. Maybe end-user experience in cloud apps. It will definitely not be SD-WAN or cloud edge.