& cplSiteName &
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
Pete Baldwin
Pete Baldwin,
User Rank: Light Beer
12/5/2012 | 5:26:52 PM
re: VMware Insists It's Not Warring With Cisco

Looking at the Dell example, and the strained Cisco/HP relationship... all these companies are going to end up competing with each other, especially in the data center. They can be friends for a while, but it seems like the pressure is going to increase over time.

VMware still doesn't have a switching component of its own, but someone on the earnings call asked about the possibility of VMware doing its own white-box kind of switch.  VMware said they're not interested, and I don't blame them. But if they got really ambitious (and really didn't mind pissing off Cisco), there are a couple of candidates that could be had.

That's looking way ahead, though. I can believe VMware and Cisco will play nice for now, and as Henlin mentioned to me, Cisco seems to be putting more into VCE (having put up Praveen as CEO for the partnership, e.g.). 

User Rank: Light Sabre
12/5/2012 | 5:26:50 PM
re: VMware Insists It's Not Warring With Cisco

Yes Craig and you have only mentioned the obvious Hardware guys.

You could also include;


....and what about the software players? Think;


As the network converges and becomes a utility, the competive distinction will blur further and they will all have to compete for the same consumer &amp; enterprise purse, albeit with different niche requirements &amp; demographic needs.

User Rank: Light Sabre
12/5/2012 | 5:26:49 PM
re: VMware Insists It's Not Warring With Cisco

So what happens next?

Will EMC (which owns 80% of VMware) buy a networking hardware company? Or partner with all the whitebox hardware vendors and/or the "ABCs" ("Anything But Cisco")?


User Rank: Light Beer
12/5/2012 | 5:26:48 PM
re: VMware Insists It's Not Warring With Cisco

In 2000, the runtime performance demands of most business software application were not pushing the limits of performance achievable with sequential execution. A single standard CPU could host dozen(s) of VMs. The processor clock rates stopped increasing in ~2005, but the application performance requirements have kept increasing. Now VM software supports (technically) multiple CPU cores under the same VM, but the VM abstraction does not help in parallelizing the application execution over multiple cores. Other than for embarrassingly parallel programs, it often is very hard to get performance gain by adding more than just a handful or so parallel cores. But when one does need to keep up with the realtime performance requirements, eg the ever increasing analytics event rates, data volumes and complexity, doing increasingly finer grade parallel processing becomes necessary -- often requiring to remove any intermediate software abstraction layers such as VMs. How is VMware addressing this emerging customer need?

In networking, an increasing number of applications with demands for deterministic (and high-end) latency, jitter, throughput, reliability and/or security guarantees, need connectivity below the packet-layer shared L2/3. Does the SDN virtualization go down to L1/0, without being constrained by the (conventionally very strict) functional limitations of the 'white box' hardware being programmed? Sure, one can virtualize shared L2/3 networks, but multi-client shared L2/3 switches unavoidably cause non-deterministic latencies and bandwidth availabilities, as everything affects everything in these network architectures. Going forward, an increasing portion of high-revenue network applications cannot tolerate the uncertainties of multi-client shared L2/3 switched/routed networks.

Non-vendor specific abstractions, such as VMs are essential for productivity, flexibility and manageability. At the same time, there is a need to provide improved, and deterministic performance. Software abstractions work against this objective. Moreover, "software defined networks" are in reality still hardware limited networks. Can't program in software 'white box' hardware equipment to do what they don't support. For high-performance and high-efficiency networking, as well as computing, we do need innovative capabilities beneath the traditional software layers. Instead of additional layers of abstraction/virtualization software over generic 'dumb' hardware, the new hardware needs to provide a higher level of programming interface to software, to provide improved per performance and efficiency.

Yes, the functionalities need to be definable in software from the management systems without vendor-specific idiosyncrasies. But the key necessary new functionalities and capabilities need to be actually enabled by new hardware, so that they exist in the first place and can then be configured in software, in an open standard manner, by the system operator and/or service users.

VM software abstractions have worked fine in the past. The application performance requirements mostly have not exceed the limit underlying hardware capabilities in the past, but they will increasingly going forward. Of course one should pursue the productivity, flexibility and manageability benefits by the easiest way available that still meets the performance requirement. Software abstractions such as VMs have fit this purpose fine, until recently for the more latency critical applications -- and this trend will continue among greater and greater number of business applications. Virtualizing networks using the same decade+ year old software abstraction models without architectural innovations all the way down to hardware is ignoring the fact that application performance requirements will be increasingly exceeding the capabilities of present gear of hardware (in computing, limited by the design for sequential execution, and in networking limited by data load adaptive bandwidth allocation stopping at L2, causing hugely inefficient, though expensive, high nominal bandwidth capacity, physical layers).

User Rank: Light Beer
12/5/2012 | 5:26:46 PM
re: VMware Insists It's Not Warring With Cisco

The analogy between what VMware does with computer hardware and what SDN aims to do with router hardware is an interesting one.

By virtualizing computer resources VMware has fueled the growth of the cloud market. But I'm hearing more and more that the network hasn't kept up and people increasingly are questioning what value there is in being able to add computing resources on the fly if they can't do the same with the connection to the data center.

Seems like this acquisition is VMware's attempt to address that.


Featured Video
Upcoming Live Events
September 17-19, 2019, Dallas, Texas
October 1-2, 2019, New Orleans, Louisiana
October 10, 2019, New York, New York
October 22, 2019, Los Angeles, CA
November 5, 2019, London, England
November 7, 2019, London, UK
November 14, 2019, Maritim Hotel, Berlin
December 3-5, 2019, Vienna, Austria
December 3, 2019, New York, New York
March 16-18, 2020, Embassy Suites, Denver, Colorado
May 18-20, 2020, Irving Convention Center, Dallas, TX
All Upcoming Live Events