2013 IDC Directions Part II- New Data Center Dynamics and Requirements

March 17, 2013
By

Introduction: #

In this second article on the IDC Directions 2013 Conference (March 5th in Santa Clara, CA), we cover the two sessions that addressed the changing dynamics of the new Data Center (DC). We examine the many challenges and requirements new DCs must address, especially for geographically distributed sites. Next we look at current trends, projected growth rates and emerging technologies, such as Software Defined Networking (SDN), and the role they might play in the DC of the future. Finally, we look at why converged DC infrastructure will be a huge growth opportunity till at least 2016.

Note: Please see 2013 IDC Directions Part I  for an explanation of the “3rd Platform” and its critical importance to the IT industry.

Mega Datacenters and Content Depots: The New Physics of IT in the 3rd Platform Universe: #

Sr. IDC Analyst Rick Villars asked a rhetorical question to start his presentation: Are Datacenters up to the task? Evidently not! IDC found that 84% of DCs had issues with power dissipation, cooling, space, capacity, and up-time, which negatively impacted business operations.  The solution was said to be DC expansion- in three different ways:

  1. Global expansion: deploy DCs in new geographical regions to provide business services to more clients in different locations.
  2. Offer new services and applications
  3. Massive Increase in scale: DCs need to be able to accommodate more users, different types of applications and much more data to be processed, stored, and retrieved.

Three key Data Center issues were said to be:

  1. Types of DC workloads, e.g. transactions, computing, content servicing, archiving, analytics
  2. Variable computing, i.e. variable workloads based on type
  3. Data gravity-physical placement of DCs will be shaped by available network connectivity, power consumption, cost of land and application characteristics NOT necessarily the DC provider’s business location.

There’s a growing importance of “data factories” based on hyperscale and hyper-standardization. Some companies are expanding the type of DC workloads they’ll handle. For example, Amazon has traditionally offered content and archiving, but is now adding compute and transaction capabilities to its DCs.

DC agility is a never-ending journey. We’ve migrated from traditional batch processing to virtualized servers to converged infrastructure (server, storage and networking). This has resulted in much quicker service/application deployment times- from months to days to hours. Converged DC infrastructures will grow at a 40% CAGR to $17.8B in 2016 vs <2% CAGR for non-converged (i.e. separate) DC infrastructures.

SDN [Software Defined Networking] may provide new provisioning and management services that could result in even faster application deployment times (e.g. maybe in minutes). SDN may be the technology of choice to inter-connect DCs, according to Mr. Villars. Here are the other key points he made during his presentation:

Impact of Variable Computing:

  • 10% of IT server assets will handle 50% of workloads
  • Solutions that boost IT agility for variable loads will dominate the management agenda
  • Radical changes ahead for application software licensing and life-cycle management
  • Competing in public cloud is all about maximizing revenue from variable computing

Impact of Data Gravity:

  • Growing distinction between data owners and data custodians
  • Cloud providers have or will become the new leaders shaping storage technology choices
  • Most new DCs will be located close to cloud building
  • Operating DCs can’t be a part-time job
  • SDN discussion shifts from intra-DC to inter-DC (connecting DCs)

The Future of The Data Center: it’s not just a place to keep computers and storage servers

  • The first point of contact with customers
  • The foundation for new business model(s)
  • Maintaining a DC is not a part-time job! “3rd Platform” IT environments change all the rules for capacity planning, which becomes much more complex and time-consuming.
  • Manufactured, not constructed, to provide the best services, agility and scale
  • The DC is the system and is the business of the provider (whether it’s used for internal or external customers)

Why the Datacenter of the Future Will Leverage a Converged Infrastructure: #

IDC Group VP Matt Eastwood said that the “3rd Platform” Requires a Different Type of Infrastructure. The “re-aggregated” DC infrastructure should be an “Enterprise DC” with the following characteristics:

  • System Level performance
  • Legacy
  • Heterogeneous equipment/services
  • Bladed/Converged
  • OPEX optimized
  • Support a complex portfolio of applications

Image courtesy of IDC.

Image courtesy of IDC.

In addition, the “3rd Platform”is placing increased burdens on the DC to Scale in terms of number of users, explosion in data, and more services/applications. This requires better programmability, more agility, increased need for internal analytics, better systems management in public/private/hybrid cloud environments.

70% of DC apps have now been virtualized. The next step is a converged infrastructure with OPEX reductions overtaking CAPEX as the main cost driver. Such a converged will provide flexibility, faster time to market for new services, reduced incidents of down-time by 50 to 75%. It’s fundamental characteristics include :

  • Integrated server/storage/network/management
  • General-purpose distributed workloads
  • Single vendor sale and support
  • Single SKU/complete system/support

Image courtesy of IDC.

Image courtesy of IDC.

IDC estimates the converged infrastructure DC will reduce cost per user by 50% each for server, network, power and cooling; and by 25% each for storage and facilities.

IDC says the worldwide converged system market was only 3.9% of DC spending in 2012, but will grow to 12.8% of all DC spending by 2016. The firm forecasts 2016 CAGRs of 42.7% for software, 37.1% for storage, 40.9% for servers, and 40.1% for the (internal DC) network. Cisco was seen to be the big winner with a 17.4% increase in market share by 2016. Dell was forecast to be in second place with a 5.7% market share increase by 2016.

Conclusions: #

Image courtesy of IDC.

Image courtesy of IDC.

The explosion in applications, data, devices and users challenges CAPEX and OPEX in traditional IT driving need for an integrated 3rd Platform Data Center which will result in:

  • Lower Costs: Reduce DC floorspace, power and cooling, capital spend and operational requirements
  • Speed: Market forces drive need for faster time-to-market
  • New Apps: Users see opportunity to do something different
  • Flexibility: Need to efficiently automate and virtualize data centers requiring pre-tuned infrastructures

Looking Ahead:

The third article in this series on the 2013 IDC Directions conference will take a hard look at SDN as discussed in the session: Evolution or Disruption: Where Are We Headed with Software-Defined Networking (SDN)?  We will also examine why it’s very unlikely SDN will be used within a DC, but has better potential for inter-connecting DCs.

—————————————————————————————————————

For an in depth look at Data Center Dynamics from both a silicon and equipment vendor perspective, please attend the April 10th  IEEE ComSoc SCV meeting in Santa Clara, CA.

Meeting logistics and RSVP information is at:  http://www.ewh.ieee.org/r6/scv/comsoc/index.php#current

Tags: , , ,

3 Responses to 2013 IDC Directions Part II- New Data Center Dynamics and Requirements

  1. Anonymous
    March 18, 2013 at 1:37 pm

    Excellent summary providing keen insight and IDC perspective of the evolving Data Center. Was there any discussion of multi-media content and requirement to bound end-to-end transit delay and jitter? Thanks in advance!

  2. Yigang Cai
    March 18, 2013 at 9:31 am

    Alan,

    Thanks for this sharing. Your article on DC is very informative, nice summary. I will share it with my colleagues.

    Not only IT, telecommunication world has datacenter problem too. For example, we are developing a PCMD (Per Call Measurement Data) for Verizon LTE network. PCMD generate huge data in real-time. Last month, we invited a professor Diego Klanjan of Northwestern University giving a talk on “Big Data Today and Tomorrow”. Next month, we will organize a seminar on “Telecom Data Analytics”.

  3. March 17, 2013 at 9:43 pm

    Thanks Alan for summarizing.

    What the difference between a data owner and a data custodian, that is referenced above? I am thinking the the data owner would be in the context of a private cloud where the data center owner also owns the data. Whereas the data custodian would be the situation where a 3rd party data center is hosting data.

    The idea that the data center can be the first point of contact for customers could portend good things for independent operators with C.O.s or headends that could be turned into a data center.

    As an example, I know of one independent operator is treating their data center as a disaster recovery center for their data center customers. This provides added value that help them differentiate from larger, non-local, larger competitors that might have lower costs.

Leave a Reply

Viodi Tweets

%d bloggers like this: