It’s been a busy week here at Plexxi. On Tuesday, we announced our partnership with Cari.net, a high-performance, scalable and flexible hosting platform based on Microsoft Cloud OS. CARI.net’s newly released CARIcloud service is powered by Plexxi and uses software-defined networking to allow companies to automatically adjust to conditions on their networks and make sure that the most important applications are never starved for performance. The platform enables customers to manage organizations and scale their data centers without being restricted to a single cloud service provider.
In this week’s PlexxiTube of the week, Dan Backman explains how Plexxi’s datacenter fabric transport solution is different from a more traditional WAN gateway.
Art Cole contributed an interesting piece to Enterprise Networking Planet this week on customizing IT hardware in a “software-driven” universe. In my opinion, we tend to think about the discrete layers within information technology hardware—the boxes that make up the network, the servers that make up compute, and the devices that make up storage. Having flexibility in each layer of hardware is crucial, but we also want the same flexibility in the interconnect that ties them all together. We want programmability and dynamism elsewhere in the stack for the same reasons we strive for dynamism in the interconnect. A bunch of dynamic things statically connected doesn’t offer the true unbounded freedom that the future of applications requires – particularly as we see the growth of Big Data. Look for technologies popular on the transport side (WDM, for example) to make a splash in the data center.
In an article for V3.co.uk, technology editor Daniel Robinson spoke to how Big Data is reshaping the landscape of enterprise IT and the challenges that it is bringing to the table. It’s worth pointing out that the challenges he identified exist across all of compute, storage, and networking. While compute and storage have embraced scale-out architectures for years, networking is a slightly different beast. Replication that distributes and backs-up data puts a lot of load on the network. This is why a lot of new Big Data deployments frequently come with a new network to support it. The fear is that mixing environments (Big Data and residual) will put one or the other at risk if anything goes wrong. Ultimately, businesses looking to move beyond Big Data experimentation will find that the network needs a bit more attention than it has had to date. It can be prohibitively expensive and time consuming to try to swap out architectures half way through.
Mary Shacklett cites Big Data, Wide Area Network (WAN) optimization and DevOps as major areas IT professionals should focus on when preparing 2015 budgets in a recent article for TechRepublic. In addition to the areas Mary mentions, I suspect data security becomes more important over time as well. Well-publicized breaches are putting additional pressure on an already hot topic. For enterprises, they have to have measures in place to protect data and to monitor when that data is used (Netskope is an interesting company to watch here). DevOps and automation both speak to how architects will manage data in compute, storage and networking. The resulting spend associated with this shift might be captured by one or both of those items. Also, the increasing use of each of these is worth calling out as it impacts what gets funded and who is involved.
In an article for SearchSDN this week, Chuck Black defines what ‘open’ means in software-defined networking. The discussion on open is confusing largely because being open for the sake of being open is not actually a goal. Typically, people use open to mean one of five different things: interoperable, interchangeable, standard, open access or open source. While people gravitate towards standards, the real business objective is usually more about interchangeability and interoperability. You can be standard and not be interchangeable (people choose to implement standards in often fantastically unique ways). The point is that the real conversation is not about open but about something more precise. Without that precision (the business objective behind the request), the discussion isn’t terribly meaningful.