A series of blogs began a speculation about the impact of virtualization on the network.  It’s certainly fair to suggest that the network has had little impact on first stage VLAN virtualization, or virtualization-lite.  The real question, however, is whether or not virtualization (or VMotion specifically) will stay contained within ever denser VLANS.

F5 Networks MacVittie blogged about it recently, suggesting that VM density was becoming the standard measure of IT efficiency.  We all know how that story will end.

At Infoblox we were already planning a webinar with Nemertes, Cisco and VMware for late September on virtualization and the network, so this topic was of natural interest.  We don’t think you’ll be disappointed given the lineup of experts which Andreas Antonopoulos has assembled: Chris Hoff, Mark Thiele and Richard Kagan.

Read more...

A group of bloggers are exploring the implications of virtualization for the enterprise network as network experts meet at SRI to explore the potential of infrastructure 2.0.

Read more...

You’re going to need a dynamic infrastructure lest you effectively negate the gains achieved by higher VM densities

In the continuing saga of “do more with less” comes a new phrase that’s being tossed around: VM density. For example, VMware puts forth the notion that the Total Cost of Ownership (TCO) of virtualization technology must consider VM density, saying, “Density matters in a many-to-one relationship.” VMware illustrates this concept in the context of TCO, but in general an increasing number of solutions are beginning to tout not only the benefits of higher VM density, but of their solutions ability to affect it. They recognize it as a valuable measure of efficiency. The general measure is this: the more virtual machines per physical server, the better. 

VM density relates closely to IT efficiency in just about any way you measure it: whether it’s the “cost per application” or “cost per user”, the density of virtual machines per server is going to factor into the equation. But so does the cost of managing those virtual machines, and the flexibility inherent in virtualization and cloud computing models comes at a cost: managing an increasing complex network and application network manually.

THE MATHEMATICAL DILEMMA

The dilemma is that the cost of management increases across IT as VM density grows, which actually decreases efficiency – at least as a measure of cost per virtual machine. IDC predicts not only the increase in VM density – as does Gartner and just about every other analyst firm – but also that the ratio of administrators to virtual servers will increase accordingly. By 2012 IDC predicts in its “Data Center of the Future (March 2009)” that the typical 25:1 VM to administrator ratio will increase to 35:1.

Unfortunately, the management costs incurred by a virtual machine are not decreasing despite the math that says it should. In fact, the increasing complexity posed by the growth of virtual machines will have a deleterious affect on the cost of managing the overall infrastructure, especially at the network layer. That’s because there’s more involved in this type of change than just movement of a virtual machine. The underpinnings of the network: DHCP, DNS, IPAM, are all impacted – and not necessarily favorably – by the increase in VM density.

Greg Ness says it well in “Today’s Networks Resemble Yesterday’s Factories”:

“This growing tension between system automation, increasing VLAN density/VMsprawl and rising network manual labor costs sets the stage for massive network innovation in management/automation, security and application delivery. Without innovation the network becomes more expensive to operate and less relevant to the ongoing march of IT.”  

NEEDED INNOVATION IS FOUNDED ON INFRASTRUCTURE 2.0

Infrastructure 2.0 offers the foundation on which solutions can be constructed to counter the costs of higher VM densities and increasing management expense. Whether it’s network, application delivery network, or at the IP address management layer, the ability of solutions to collaborate via integration and take decisive actions based on the data exchanged between such integrated infrastructure is what is required to create the data center of the future that isn’t so prohibitively expensive to run that it never comes to fruition. It is the glue that will hold together the data center in the future.

The analogy of today’s network as a factory is altogether too true; packets come in one end and out the other comes a response. It’s an assembly line, with every packet and request being treated as equally as others. Customization is not possible, and any change to the process is disruptive. Along comes dynamic infrastructure; an infrastructure that takes as a core precept the concept of context. Context gives infrastructure the ability to not just deal with the change inherent in high density virtualized architectures, but act upon it in a flexible way. When an infrastructure is dynamic and context-aware, it can customize and treat each request as an individual entity deserving of the “special attention” once only afforded expensive, custom processing. And it is that flexibility coupled with programmability from whence innovation comes.

While VM density may be the new measure of IT efficiency, part of that metric must include the cost to manage the increasing rate of change it brings with it. In order to keep those costs down it will be necessary to architect a network and application network that takes advantage of the integration capabilities of Infrastructure 2.0 while leveraging the flexibility inherent in this new breed of infrastructure solutions.

Many of today’s large enterprise networks are burning cash and are overly dependent upon layers of manual processes managed by legions of clerks to stay available and secure.  Networks were created this way from the start; they were architected to work just like the environments they were about to transform.

From those early days until today they have replaced populations of “middle men” and paperwork with streaming electrons racing between larger populations of endpoints.  They drove incredible IT innovation and productivity gains.

Network vendors, meantime, had little incentive to do more than to produce ever more powerful network appliances requiring ever greater legions of tech laborers with their own budgets and buying privileges.  The productivity gains were so enormous that no one cared nearly as much about productivity within the network itself.

Those gains drove incredible wealth creation in the Valley and powered the fates of companies like Cisco, Juniper Networks, Extreme Networks, Foundry, Brocade and Riverbed.  It helped to foster massive investments in venture capital that ultimately drove even more innovation.

Yet today these companies are facing a potential sea change in how IT services are managed and delivered.  And that is a problem.  Ironically, today’s networks are often run like yesterday’s factories. 

Read more...

Virtualization will reinvent the IT world... when networks are ready.

Read more...