Richard Napolitano

Subscribe to Richard Napolitano: eMailAlertsEmail Alerts
Get Richard Napolitano: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn

Related Topics: Cloud Computing, Big Data on Ulitzer, DevOps Journal

Blog Post

The Next Era of IT | @CloudExpo #Cloud #DevOps #BigData

An agile and dynamic network to keep pace with applications, compute & storage

I have been on the road a lot lately meeting with customers, prospects, partners and investors and it has become clear to me that there is a very simple, but profound trend emerging in Information Technology. Today you hear a lot about Third Platform, Bimodal IT, Hyperconvergence and Software Defined ”X” (Networking, Storage, Data Center etc.). The discussions about these technologies are prevalent, but the question is why? Why these topics and why now? Why should we care about Third Platform, Bimodal IT, Hyperconvergence and SD”X”? The answer is simple; storage and compute have evolved to be highly dynamic, and the network itself now needs to keep pace.

Looking Back
Over the last two decades, data center architectures have traditionally been highly static from the perspective of both the application and the network. Designing a data center was a lengthy process of capacity planning and design, after which the network team built a data center network around physically redundant two-tier leaf/spine architectures. Depending on pre-planned growth models and expected product lifecycles, multiple chassis were typically used as the spine. Leaf switches, dual homed back to the spine, were mounted at top of rack (ToR). The network, server racks and rows all got cabled together, were tested and deployed. That design created a static infrastructure, defined by the physical cabling, but it was optimized for the north/south traffic that occurred from the applications to the end users. As new applications were needed to support the business, the server team would build each discreet application on it’s own infrastructure and the network team would then configure VLAN, routing, security, and quality of service parameters to make the application live. The infrastructure was rigid, defined by cabling, and applications were “poured” into it.   This was a manual process, accomplished with separate siloed teams, which typically took a few weeks from start to finish, but life was good as this process worked.

With the advent of server virtualization and the emergence of VMware/ESX, HyperV and OpenStack, application silos were broken down and many different applications were converged on a common, yet still static, compute infrastructure. The network architecture remained unchanged, but the number of physical servers in the data center was significantly reduced. Instead of a 1:1 ratio of applications to servers, you saw somewhere around a 10:1 ratio.   Due to the fact that IT was not yet taking advantage of features like dynamic vMotion, these new virtual machines stayed on the same server and appeared static from the point of view of the network.

Time marched on and the number of users and applications exploded, as well as their data sets (Big Data). The server and application teams began to leverage virtual machine mobility, making the physical location of the applications variable, and generating significant east/west traffic patterns. As the server and application teams attempted to deploy these new highly virtualized and mobile applications (Hadoop/SAP HANA, VDI, etc) onto the still unchanged and static network infrastructure, a functionality gap began to form. This disparity between the rapidly changing and dynamic compute and storage domains, along with the unchanged, static network, further segregated the server, storage and network teams. This segregation led to workflow inefficiencies and increased the chance of errors occurring.

The months turned into years and the static nature of the network left IT organizations in a pinch. In an attempt to support highly virtualized and dynamic applications, the network industry created overlays, protocols, one-off fixes and other short-term solutions in an attempt to keep pace. This led to today’s complexity avalanche and with each new “work around” the network infrastructure has become even more fragile and unstable. This model is simply unsustainable.

Looking Ahead to the New IT Era
In the next Era of IT, it’s no longer viable to build a static network and “pour” applications into it; which leads me back to why the industry is making so much noise around Third Platform, Bimodal IT, Hyperconvergence and Software Defined ”X” (Networking, Storage, Data Center etc.). They are all focused on efforts to move away from manual, slow and static approaches and migrate to ones that are agile, rapid and dynamic.

In the new IT Era, you need to start with the APPLICATION and DATA in mind and then build a network with the ability to support them, not the other way around. The decades-old practice of pre-architecting, designing and implementing a static infrastructure simply won’t support the dynamic needs of the business moving forward.

In my next blog, I’ll capture some of my findings from the road about how forward thinking end users are deploying technology for the next IT Era. I welcome your comments and thoughts.

The post The Next Era of IT appeared first on Plexxi.

Read the original blog entry...

More Stories By Richard Napolitano

Successfully leading a technology company requires a strong sense of what the market needs and an uncanny ability to build it. Plexxi CEO Rich Napolitano combines the vision required to disrupt with the experience necessary to succeed. He has spent more than 30 years in the IT industry, most recently as the head of EMC’s Unified Storage Division, where he was responsible for more than $4 billion in annual revenue and 2,000 employees.

Prior to EMC, Rich has led both sales and engineering, and acted as both founder and venture capitalist across companies like Sun Microsystems, Pirus Networks, and Alchemy Angels. Rich holds a BS degree in Computer Science from the State University, Stony Brook, NY, and owns two U.S. patents.