Overview:

The Issue:

The Issue:

IIT security staff, faced with the challenge of securing the inevitable flux in
their infrastructure, are usually stuck in reactive mode. They react – to systems
upgrades, mergers, and acquisitions; to the re-centralization of most IT function
into data centers and the consolidation of data centers; and to the spread of all
sizes and kinds of organizations over ever more space as a result of the
continuing 9 to 11% growth in the number of branch offices. Proactive security –
helping plan and execute security changes to enable adoption of new tools and
technologies – falls by the wayside.

IT security is set up to prevent and react to security problems, not to set
acceptable levels of risk. Significant increases in risk are traditionally viewed as
automatically “bad”. Given the difficulty of securing the complex interfaces
among different architectures, silos, and generations of technology, optional
changes and elective complexity are resisted if not simple to secure. How then
can IT security shift from a reactive to a proactive position?

One action security teams and IT are increasingly performing to reduce
risk and manage complexity is set policies to guide ongoing operations. By
defining policy, one can lay out more secure operational modes for everyone and
make dealing with complex infrastructures less a matter of individual memory,
capacity, and preference, and more a matter of documented practice.

The Issue:

Major tectonic shifts in the way enterprises work with and provision their
core applications are forcing changes in the way the enterprise has to think about
securing them.

One shift is the continuing opening of the enterprise, with the gradual
federation and interpenetration of IT systems between an enterprise and its
partners, customers, and suppliers. The figurative walls of the data center are
being filled with doors, windows, and access ducts, and now serve more as a
framework for structuring the flow of information than as a barrier to it.
Another shift is the rise of service-oriented architectures (SOAs).

Enterprises are looking to SOA to provide an integration method for their
applications, a development methodology and framework, and an overall
architecture and philosophy for deploying new functionality. As enterprise
applications gain services interfaces, and sometimes are actually atomized and
turned into constellations of loosely-coupled services, each service creates on the
network a new set of access points; perhaps tens or hundreds of times as many as
there were before. Things that used to happen within an application, on a single
server, become network traffic among servers and even among data centers.
Some formerly internal functions even become invocations across the Internet of
software-as-a-service (SaaS) packages, or services in partner or supplier data
centers. Moreover, components in a SOA can scale independently of each other:
new instances of an application running on a Java application server might be
created to handle peak loads, and then destroyed as the load subsides.

Read this Issue Paper:

Clients:New Suit of Armor: Securing the Data Center

The Issue:

Server virtualization is one of the most-discussed technologies of the past
few years. We find that although some organizations are already generating
substantial savings with virtualization in their production environments, the
majority of participants in Nemertes’ Security and Information Protection
benchmark research are not yet using virtual servers in production. They plan to,
however, looking for the increased resource utilization, broader platform
standardization, and deeper management automation that server virtualization
enables.

As virtual servers move into production, IT needs to address security and
compliance issues. Unfortunately, most participants in the benchmark, when
asked how they secure their virtual servers, say they treat them like physical
servers as much as possible! Sensibly, they use host-based security such as antivirus
and anti-malware agents. However, they also use network tools to protect
virtual servers exactly as if they were simply very thin, very densely stacked rackmount
boxes.

The Issue:

Server virtualization is one of the most-discussed technologies of the past few years.
We find that although some organizations are already generating
substantial savings with virtualization in their production environments, the
majority of participants in Nemertes’ Security and Information Protection
benchmark research are not yet using virtual servers in production. They plan to,
however, looking for the increased resource utilization, broader platform
standardization, and deeper management automation that server virtualization
enables.

As virtual servers move into production, IT needs to address security and
compliance issues. Unfortunately, most participants in the benchmark, when
asked how they secure their virtual servers, say they treat them like physical
servers as much as possible! Sensibly, they use host-based security such as antivirus
and anti-malware agents. However, they also use network tools to protect
virtual servers exactly as if they were simply very thin, very densely stacked rackmount
boxes.

The Issue:

The history of computing shows several major architectural changes, most
of which were quite clear and easy to discern (or is that just 20-20 hindsight?).
Today it seems a lot harder to discern what the next computing architecture will
be, but in fact the trends are all pointing in the same direction.

Looking back we see that mainframe computing was partially eclipsed by
client-server computing and then n-tier web architectures. Partially eclipsed,
because no part of computing history ever disappears. Any sufficiently large
computing organization is a bit like a museum: you will find different stages of
history preserved on the pragmatic basis of “if it isn’t broken, why fix it”.
Mainframes still abound, client-server is king and n-tier web is growing in leaps
and bounds. So what is the next-generation computing architecture?

Puzzlingly, if you look at the data center today, it almost seems like we are
trying to re-create the mainframe from distributed components - on-demand
computing, provisioning and orchestration, scheduling and coordination - all
these activities seem to be attempts to build a data-center-scale mainframe. But
at the same time, computing within the data center could not be more
distributed. Far from consolidating computing onto bigger and bigger CPUs, we
seem to be moving to plug-and-play blade servers.

Sun announced that it will ship a hypervisor, xVM, and a management product, xVM Ops Center, aimed specifically at heterogeneous virtualized infrastructure. xVM combines a stripped-down version of Solaris with the Xen open-source hypervisor to provide a somewhat light-weight virtualization infrastructure that can use core Sun/Solaris technologies for storage virtualization and for avoiding hardware-based service interruptions.

The Issue:

The shift to a data-center centric model for enterprise computing
continues to increase the importance of the reliability, scalability and
management of the underlying network infrastructure. Consolidation of sites
and resources demands major changes in WAN architecture and backup
strategies. New technologies such as virtualization, storage virtualization, Web
services and clustering will dramatically affect information processing and so also
network requirements, leading not only to increasing demands for bandwidth,
but also to fundamental changes in the way data flows across enterprise
networks. New end-user applications such as voice and video create contention
issues for available bandwidth. Enterprise organizations must plan for these
changes, and must concentrate on obtaining and delivering network services
capable of meeting these changing requirements.

Organizations are, at the same time, focusing on data-center management
and operations, attempting to cut costs, improving efficiency, automating
repetitive tasks, and better aligning IT spending with business needs and demand
for services. More than three-quarters of enterprise IT executives interviewed for
Nemertes’ “New Data Center” benchmark take a holistic view of the data center
as part of a broader shift in IT culture to “IT as a service.” This translates to data
centers as business-service delivery centers, with critical requirements for
performance and availability.

Read this Issue Paper:

Clients: Network Services and the New Data Center 

The Issue:

The “data center network” is a myth. For nearly as long as there have been data centers (DC), there have been several DC networks that interact with and overlap one another, most importantly the data, storage and high performance compute (HPC) networks. The desire to consolidate these networks onto a single fabric is as old as the networks themselves. As network vendors continue to reengineer and ramp up production of 10G Ethernet equipment, the promise of unifying data, HPC and storage networks onto a common technology—Ethernet—increases. Network vendors have some significant technical and engineering hurdles to clear before they can simultaneously meet the opposing pulls of storage, which demands lossless reliability, and high performance applications, which demand very high throughput at very low latency.

Consolidation onto a single fabric—a DC over Ethernet—will reduce physical complexity, lower material costs and simplify operations. Ultimately, though, the most important benefit of a unified DC fabric will be increased enterprise IT agility deriving from the ability to rapidly and dynamically reprovision network resources across data, storage and HPC domains.

Read This Issue Paper

Clients: Data Center IO Consolidation 

Non Clients: Nemertes Issue Papers are available to clients only. If you're not a client and would like to receive a copy of the Issue Paper, please contact us.

  By Sandra Gittlen, Newsfactor.com, 04/11/07

Many companies' data center responsibilities are broken out piecemeal, but experts say that companies embracing New Data Center technologies, such as blade servers, grid computing and virtualization, will succeed by consolidating the management of all critical functions into a single role. Using these advanced technologies begs for someone capable of bringing an integrated, holistic approach to data center architecture and design, says Johna Till Johnson, Network World columnist and co-founder of Nemertes Research.

The Issue: A New World to Secure

Data centers today are truly “new” from every perspective: facilities, storage, management, computing, and networking. Although data centers have existed as long as enterprise computing itself has, a confluence of economic, enterprise, and technological changes is driving a major metamorphosis in data center design and implementation. This, in turn, is determining how data center and security professionals approach the problem of securing the data center and the enterprise network from threats, internal and external.

SLAs for hosted, managed and co-located data center services

New Data Center Strategies Newsletter, By Andreas M. Antonopoulos, Network World, 1/30/07

Almost 10% of participants in Nemertes’ data center research cited outsourcing data center facilities as one of their most important funded initiatives.

Over the last five years, the cost of building data centers has increased quite dramatically, mainly because of the increased power and cooling demands of dense platforms such as blade servers. So naturally, many businesses have decided that data centers are too complex and too costly to own, even if they are necessary to have.

Things to consider when leasing data center equipment

New Data Center Strategies Newsletter, By Andreas M. Antonopoulos, Network World, 1/23/07

Many of the IT executives I speak with lease their servers from one of the big server vendors. With a bit of careful planning they can use the lease to manage technology refreshes on a two-year basis, thereby always staying one step ahead of technology obsolescence.

Liquid cooling systems

New Data Center Strategies Newsletter, By Andreas M. Antonopoulos, Network World, 1/16/07

The upcoming final volume of Nemertes' data center research looks at the facilities challenges in the data center. Power and cooling were clearly the “hot” (pun intended) topics for this year.

We’ve discussed strategies for both power savings and cooling in this newsletter. One question we are often asked by our clients is, “Are servers going to keep getting hotter?”

Considerations for locating a data center

New Data Center Strategies Newsletter, By Andreas M. Antonopoulos, Network World, 1/2/07

Real estate mavens will endlessly repeat the three criteria for a property: location, location, location. Data centers are real estate, too, so what are the three most important criteria for the data center?

Location is the most important consideration for a new data center, as with any piece of real estate. But, what makes a good location for a home does not make a good location for a data center.

New Data Center Strategies Newsletter, By John E. Burke, Network World, 12/19/06

Could RAM be moved off server blades?

No, POV is not some hot new data-center technology that you haven't picked up on yet. It's a completely non-technical term that you might remember from freshman composition: POV means "point of view," the perspective from which a story is told. Just as in fiction, the point of view from which an IT story is told can have a dramatic impact on what it means and how we respond.

New Data Center Strategies Newsletter, By Andreas M. Antonopoulos, Network World, 12/12/06

Could RAM be moved off server blades?

Over the last five years, servers have been deconstructed into different parts, and some of those parts have ended up on a network. Servers have transformed into blades, and power supplies, network interfaces and storage are shared across many blades, which contain just the basic CPU and memory.

Business Communications Review: Four Key Trends In Building Better Datacenters
Drew Robb

Any number of experts will give you their opinions on the hot trends in datacenters these days. For example, analyst Andreas M. Antonopoulis, senior VP of Nemertes Research, lists consolidation, growth, availability and operational efficiency as the four main themes that come up in Nemertes surveys of IT executives. Others emphasize such

New Data Center Strategies Newsletter, By Andreas M. Antonopoulos, Network World, 11/28/06

Software distribution is taking an interesting turn with virtual appliances, which combine an application and underlying operating system.

If you have ever had the experience of trying out a new application, it can be daunting: hours of tweaking operating system settings and downloading libraries (dependency hell) to get to the point of even starting up the application. For demo or evaluation software it can be the kiss of death - no matter how good the installer, it is always challenging to tame a general-purpose operating system to get a smoothly running application.

New Data Center Strategies Newsletter, By Andreas M. Antonopoulos, Network World, 11/21/06

As power becomes an increasingly large component of overall data center costs, we have examined a number of strategies for saving electricity.

Recently, I wrote about the impact of virtualization on electricity consumption. In that article we outlined a strategy for generating ROI in server consolidation - consolidating workloads and switching off the remaining servers until they are needed. This approach can save money by cutting power use, but pretty soon you will need those surplus servers to support your application growth.

New Data Center Strategies Newsletter, By Andreas M. Antonopoulos, Network World, 11/14/06

For companies considering thin-client desktops the range of choices has increased quite a bit in the last few years.

From presentation servers like Citrix to desktop blades, companies can now choose how thin to make their desktops.

One interesting approach I wrote about in a previous article is to use virtual machines in blade servers and the Remote Desktop Protocol (RDP) to deliver thin desktops from the data center. Using this approach you can commingle desktop and server images on the same pool of blade servers, alternating loads based on demand. You get to make the most of the existing servers, increase flexibility and simplify your recovery process since both desktops and servers can be recovered to the same machines in another data center.

New Data Center Strategies Newsletter, Network World, By Andreas M. Antonopoulos, Network World, 11/07/06

Power consumption has become a big issue in the data center. But with all the focus on servers, are storage systems getting an easy pass?

After all, storage arrays have one big disadvantage: moving parts. While most of the power consumption is to keep the disks spinning at high RPM, more power is used to remove the heat produced by all that spinning. Storage vendors are likely to face increasing pressure to provide power usage measurements and then reduce power consumption and heat output as much as possible.

New Data Center Strategies Newsletter, Network World, By Andreas M. Antonopoulos, Network World, 10/31/06

Scientists conducting experiments on addiction have shown that when mice are allowed to self-administer narcotics, such as cocaine, by pressing a lever, they will very quickly develop addictive behavior patterns - pressing that lever repeatedly, even until death by overdose. At a recent technology conference, an IT director described a very similar behavior that immediately reminded me of the addiction studies. In this case, however, the drug of choice was a virtual machine.

New Data Center Strategies Newsletter, Network World, By Andreas M. Antonopoulos, Network World, 10/17/06

One of the strongest trends we have found in our research is consolidation. Companies are consolidating data centers and consolidating servers and storage within data centers.

With all this consolidation, you might think that new data center construction is in a slump. And you’d be wrong.

New Data Center Strategies Newsletter, Network World, By Andreas M. Antonopoulos, Network World, 10/10/06

Of every kilowatt of power entering your data center, less than 30% will do any actual “work” for you. The rest will be lost to inefficiencies in power supplies, UPSs and inside servers. And despite the recent drop in energy prices, with China’s demand for power and the tight energy market in North America, electricity is likely to be a major cost in data centers for many years to come.

Table of Contents
1 EXECUTIVE SUMMARY 4
2 THE NEW DATA CENTER 5
2.1 OVERVIEW 5
2.2 KEY THEMES 5
3 TRENDS IN NETWORK ARCHITECTURE 8
3.1 THE IMPACT OF DATA CENTER CONSOLIDATION ON WAN ARCHITECTURE 8
3.2 DATA CENTER INTERCONNECTS 10
3.3 DATA CENTER BANDWIDTH: HOW MUCH IS ENOUGH? 12
3.4 MPLS AND THE DATA CENTER 14
3.5 WAN REDUNDANCY STRATEGIES 20
3.6 THE ROLE OF THE INTERNET 22
3.7 NETWORKING INSIDE THE DATA CENTER: STORAGE NETWORKING 25
3.8 NETWORKING INSIDE THE DATA CENTER: SERVER INTERCONNECTS 28
4 CONCLUSIONS AND RECOMMENDATIONS 32
5 APPENDIX A - BENCHMARK METHODOLOGY 35
5.1 PROCESS 35

New Data Center Strategies Newsletter, Network World, By Andreas M. Antonopoulos, Network World, 10/03/06

Nemertes’ recent research on data centers has failed to find the management nirvana: The fully autonomic, adaptive, automatic, on-demand data center. Instead, IT executive participants in our research tell us they use a combination of free, open source and commercial tools to manage their data centers.

One approach to dealing with patches

New Data Center Strategies Newsletter By Andreas M. Antonopoulos, Network World, 09/19/06

Software maintenance has long been recognized as the most expensive part of the software lifecycle. Increased security threats have forced vendors to release software updates and patches far more often, exacerbating the cost of maintenance. But a bigger problem looms: The most critical servers in an organization are not getting patched because IT managers are extremely reluctant to expose those servers to frequent updates that may cause crashes or software conflicts.