March 13, 2012

Data Center LAN Redesign: Who's Doing it and Why?

The effort by enterprise IT organizations to redesign their data center LANs is getting a substantial amount of attention in the trade press. Just how widespread is the push toward redesign, really, and what factors are driving it?

Some recent Webtorials research provides some insight into the answer. A survey conducted for the Webtorials 2011 Cloud Networking Report asked 207 IT professionals whether their organizations had already redesigned, or expected to redesign in the next year, their data center LANs in order to support cloud computing in general and virtualized servers in particular. Table 1 summarizes their responses.

We can conclude from this data that the majority of IT organizations have already begun the process of redesigning their data center LANs and that supporting server virtualization is a key factor driving that trend. However, as the same research report points out, support for server virtualization is far from the only factor driving LAN redesign. Another survey question asked that respondents indicate which of a broad range of factors was the primary one driving LAN redesign in their data centers. Their responses appear in Table 2.

Despite all the current discussion about energy consumption, Table 2 shows that reducing power requirements in the data center is not a primary factor driving IT organizations toward LAN redesign. Why not? Because, compared to the amount of power consumed by servers, the amount consumed by LAN switches is minimal. Furthermore, several other factors frequently reported on in the trade press - such as reducing complexity and increasing security - also failed to rank as important drivers of data center LAN redesign.

It's All About Cost...Or Is It?

Reducing cost emerges as the number-one factor driving IT organizations to redesign their data center LANs. This isn't surprising, as cost reduction is typically an important driver of any change in IT. What seems more unexpected is that support for server virtualization didn't make it to the top of the list. Then again, that could be misleading, as server virtualization is a contributing element of every one of the top factors on the list. For example, a primary reason IT organizations implement server virtualization is to reduce cost, both capital expenses (CapEx) and operating expenses (OpEx).

Server virtualization is also a key element in helping IT organizations implement a dynamic data center. And the fact that an increasing number of data center components, such as servers, are virtualized leads to the need to better orchestrate those components to ensure that appropriate resources are dynamically allocated and de-allocated as needed. As for scalability, it is not uncommon for a Global 2000 company's data center to have 1,000 to 2,000 servers. As IT organizations look forward a few years to where the vast majority of those servers are virtualized and each supports 10 to 20 virtual machines (VMs), the total number of VMs that must be supported could well rise into the tens of thousands.

The Path Forward

Without a doubt there is wide interest on the part of IT organizations in redesigning their data center LANs, primarily to lower costs and enable the adoption of a dynamic, virtual data center. However, the types of designs IT organizations will actually implement - and the technologies and vendors they will use to implement them - remains uncertain.


I am surprised you did not mention 10G.

I view getting to 10G copper as an imperative to reducing cost.

In my opinion, apart from cost and the usual drivers ('scalability', 'ease of management' ...), there are four main driving factors to a next-Gen DC : (1) requirement for high-density non-blocking 10G access at TOR or EOR (this is too expensive on legacy, classical switches),
(2) drive to LAN/SAN convergence on Unified Fabric
(3) Virtualization and workload mobility, bringing the inter-VM switching back to the edge of the Phyiscal network, that also self-adapts to the VMs as they appear, move and disappear and
(4) simplification towards a plug&play fabric switching system, where many physical switches act and are seen as one logical switch over any arbitrary topology.
Trying to map this to table 2,
(1) is not in, unless 10G would fall under reason #2 (scalability). It is unclear whether scalability means here : more connections, or more bandwidth + switching capacity, or both. 10G is likely to REDUCE the #of connections as compared to the current multi-1G complex constructions. Intstead, it will require non-blocking DCN designs, meaning more switching capacity and BW.
(2) seems to be captured by reason #7 from table 2 and ranks very low. It seems LAN/SAN convergence does not gain traction in accordance to its hype. *** IS THIS A FIRM CONCLUSION THAT WE MAY DRAW FROM THE FINDINGS OF THIS ARTICLE???? ****
(3) is covered by reason #4 and #5 (dynamic, virtualized) and scores high (25% on aggregate),
(4) might be mapped to #3 and #6 (reducing complexity directly aims to facilitate management) and scores relatively high, 18% on aggregate.
I started a couple of days ago an initiative within my organization to find out reasons why organizations DON'T move to a next-gen DC ... I observe heaps of literature on next-GEN DC's and at the same time, organizations seem to continue to design, buy and implement in a legacy way. Why?

Nice comment from Jan at IBM. This should take you to the comment, which had an unintentionally delayed approval.

All the factors shown in table 2 are correct and stands good if we talk about redesign. As the technology is changing and advancements are coming day by day, all such propels the enterprizes to change and revamp their LAN Network and Datacentre. This is especially true with the company’s who are having network and server room ( earlier datacenters are called server rooms) 10 years or 8 years older. Most of networks at that time are on CAT5 Standards which is giving only 100Mhz.

Available technology like CAT6 and CAT6a which is giving 250Mhz and 500 Mhz respectively adding fuel to change or redesign the LAN Network besides table 2 factors.

Applications in these dates are bandwidth hungry and require the robust medium .

It is true that cost optimization is a big factor but revamping the old network with current standards is a major point for the business to change. Once they change they advocate the table 2 factors also in gaining the budget from Finance.

I hope you understand what I mean to say.

Since we often see references to Mbps and Mhz, I asked Manish to offer a bit more information.  He kindly provided this extra information - with lots of great hyperlinks:

Category 5 cable (Cat 5) is a twisted pair cable for carrying signals. This type of cable is used in structured cabling for computer networks such as Ethernet. It is also used to carry other signals such as telephony and video. Category 5 has been superseded by the Category 5e (enhanced) specification.

The specification for Category 5 cable was defined in ANSI/TIA/EIA-568-A, with clarification in TSB-95. These documents specified performance characteristics and test requirements for frequencies of up to 100 MHz.

Category 6 cable, commonly referred to as Cat 6, is a cable standard for Gigabit Ethernet and other network physical layers that isbackward compatible with the Category 5/5e and Category 3 cable standards. Compared with Cat 5 and Cat 5e, Cat 6 features more stringent specifications for crosstalk and system noise. The cable standard provides performance of up to 250 MHz and is suitable for10BASE-T100BASE-TX (Fast Ethernet), 1000BASE-T/1000BASE-TX (Gigabit Ethernet) and 10GBASE-T (10-Gigabit Ethernet).

Whereas Category 6 cable has a reduced maximum length when used for 10GBASE-T; Category 6a cableor Augmented Category 6, is characterized to 500 MHz and has improved alien crosstalk characteristics, allowing 10GBASE-T to be run for the same distance as previous protocols.

Get E-News and Notices via Email




I accept Webtorials' Terms and Conditions.

Trending Discussions

Featured Sponsor Microsites

Recent Tweets



Please note: By downloading this information, you acknowledge that the sponsor(s) of this information may contact you, providing that they give you the option of opting out of further communications from them concerning this information.  Also, by your downloading this information, you agree that the information is for your personal use only and that this information may not be retransmitted to others or reposted on another web site.  Please encourage colleagues to download their own copy after registering at  Continuing past this point indicates your acceptance of our terms of use as specified at Terms of Use.

Webtorial® is a registered servicemark of Distributed Networking Associates. The Webtorial logo is a servicemark of Distributed Networking Associates. Copyright 1999-2015, Distributed Networking Associates, Inc.