x
Optical/IP

IDCs May Lighten the Gloom

The past few months may have been murder on the high-tech industry, but they’ve been a godsend for the sober-suited experts all too happy to contribute their commentaries on the death of the New Economy. Even the traffic reporter on my local news station has gotten off a few dot-gone quips during morning drive-time.

While these naysayers sound off, however, one part of the optical network is gathering strength and even emerging as a quiet savior of sorts: the Internet Data Center (IDC).

What is an IDC? Many things, starting with a well-built, well-powered, well-secured hosting center for Web servers, storage servers, routers, switches, transport gear, and a variety of specialized systems that add service creation and aggregation to packet networks.

But an IDC is more than the sum of its parts. I consider it the key infrastructure that will underpin a variety of businesses, most beginning with the lucrative term “managed.” Think managed hosting, managed storage, managed security, and managed applications.

If this were just about dotcoms there wouldn’t be a business model worth discussing. What’s got me going is a significant trend among enterprise customers to start outsourcing more difficult IT-related tasks. And IDCs are loaded with talented staff whose job is to manage and maintain all that hardware and software I just mentioned. Those savvy technology professionals are a scarce resource in corporate America. Ideally, IDCs give companies a way to improve their IT without having to increase staff, hire consultants, or expand internal network operations.

IDCs aren’t just a real business. They’re inextricably tied to the transformation of telecommunications from circuits to packets — and to the move from low-cost Internet services to high-value IP services.

An IDC is distinctly different from a metro POP or any other carrier metro presence meant simply as a point of aggregation or service interconnection. Companies that provide neutral colocation (think Colo.com, Equinix Inc., and the like) deliver meaningful services in metro markets, but they have a limited ability to add value to their services . Further, they may be overly exposed to the CLEC (competitive local exchange carrier) and ISP (Internet service provider) markets, which are shrinking as lenders stop lending.

An IDC, in turn, can be seen as home to a profitable series of businesses. Rather than being tied to any one market, it’s linked to the entire universe of packet-based services, which is undeniably huge, economic downturn or no.

Look at the carriers building IDCs; these aren’t second-tier players or the latest batch of private wannabes: AT&T Corp. (NYSE: T), Exodus Communications Inc. (Nasdaq: EXDS), Genuity Inc. (Nasdaq: GENU), IBM Global Services (NYSE: IBM), Intel Corp. (Nasdaq: INTC), Level 3 Communications Inc. (Nasdaq: LVLT), Metromedia Fiber Network Inc. (MFN) (Nasdaq: MFNX) with AboveNet Communications Inc., Qwest Communications International Corp. (NYSE: Q), Sprint Corp. (NYSE: FON), and WorldCom Inc. (Nasdaq: WCOM), which is also pursuing an acquisition of Digex, though a rather poorly negotiated one. Heavy hitters every one, and among the most viable carriers in the industry today. (Among the leaders, Exodus is a pure-play hosting company, and an awfully successful one, which tells me it’s a likely acquisition target in the near future.)

There are roughly 22 million square feet currently dedicated to data centers, with an additional 150 million predicted over the next five years. That suggests a possible glut, given today’s economy, but doesn’t necessarily reflect the real opportunities of the IDC market. M&A activity will probably increase in the coming years, and a core group will doubtless emerge (probably Genuity, Qwest, and WorldCom, assuming one of them buys Exodus). Many other carriers will operate data centers, since hosting and content delivery will likely become an integral part of every carrier’s business over time.

Though IDCs are built to handle just about anything that comes their way, they have three primary roles: Web hosting, storage hosting, and security.

Web hosting kicked off the whole data center market and was originally driven by the emergence of e-commerce and dotcoms. Exodus, among others, made this a serious business. Now enterprise customers are picking up the slack from dying dotcoms as corporate Websites become a critical part of mainstream businesses.

Web hosting is typically a managed service, so service providers have ample opportunity to differentiate their offerings in terms of reliability, security, support, and maintenance. The key to success lies in reducing operating costs while continually increasing the number of billable services available to customers. Reducing costs presents an opportunity for systems vendors focusing on the IDC market, since creating a complete service currently involves numerous discrete systems. This multiplicity slows provisioning, limits scalability, and keeps prices high.

Storage hosting is a newer application, but it holds equal if not greater promise. The key here isn’t simply offsite housing of servers but true management of those resources.

Basic storage hosting applications include maintenance and management, disaster recovery, on-demand capacity, and national or global database access. Storage hosting is often offered part and parcel with Web hosting, since the servers used to build Websites need their own storage facilities. Thus, a Web-hosting service typically needs a SAN (storage-area network) to support Web content. Corporate databases and other storage assets are often much too large for companies to handle, creating a lasting market for offsite storage management and a real opportunity for service providers like StorageNetworks Inc. (Nasdaq: STOR). That in turn creates significant equipment markets for makers of SAN equipment and metro optical systems that support storage protocols, such as Fibre Channel.

Security has always been the gating factor to the growth of any outsourcing business. If a company is going to move more than its home page and email out of the building, it’s going to need serious security. Firewalls are just the beginning. What’s more, since software represents a major component of any security solution, it needs to be managed and upgraded routinely. This spells “managed service” and is yet another nice play for IDC operators.

Firewalls today represent only 5 percent of hosting revenues, but they will be added to all other services, increasing their associated revenue. In addition, since IDC operators will be carrying much of an enterprise’s IP traffic, virtual private networks (VPNs) will become a core service. This also creates opportunities for service-aware switch vendors with strong security capabilities that can be used to create a packaged security service as part of a total IP offering.

What’s next? The two items to watch the most closely are application hosting and content distribution.

Application hosting is the foundation of the ASP (application service provider) market, where companies like Corio Inc. (Nasdaq: CRIO), NaviSite (Nasdaq: NAVI), and USinternetworking Inc. (USi) (Nasdaq: USIX) have staked their fortunes. Application hosting relies on a network architecture similar to Web hosting, with application software allowing enterprise customers to simplify IT management. ASPs have had to weather growing pains in the past few years, but the market is stretching out nicely now that more robust metro networks are being deployed and more intelligent edge routing systems developed.

Content distribution is another beast entirely (and worthy of a separate study). The basic premise is that the Internet is too damn slow and inefficient; if it is going to support streaming applications and low-latency access, it will require a mix of local caching and intelligent content delivery mechanisms to guarantee proper throughput.

The need for integration inside the data center is clear: The standalone routers, switches, firewalls, load-balancers, cache servers, application servers, and Web servers required to create a complete service offering make it very difficult to provision services quickly or change them in anything like real time. All that equipment also consumes a lot of power and costs a lot of money when devoted to individual subscribers, driving up service costs.

Integrated systems ideally streamline service creation and delivery while saving rack space and reducing power requirements. They also offer a single user interface, with unified management. We’ve heard this before in the MSPP (multiservice Sonet provisioning platform) market, but in this case I think it will stick because there is no real heritage to displace in the IDC. It’s all new territory, and space is a premium.

Integrated data center systems are on the way, although they’re so new that no one has come up with a meaningful moniker. Nexsi Corp. calls its box a Content Services Aggregation System, which is quite a mouthful. It's also quite a box, integrating a suite of security and Layers 4-7 networking services into a unified system. Inkra Networks is another vendor edging into this space, although it hasn’t laid claim to a name for its product. Another, Desana Systems, is building an IP service delivery system customized for the data center with ASPs in mind. There are a few stealth startups raising money right now, leveraging their expertise in Layers 4 through 7 switching, packet filtering, and content delivery. Data Center stalwarts like Cisco Systems Inc. (Nasdaq: CSCO), Extreme Networks Inc. (Nasdaq: EXTR), Nortel Networks Corp. (NYSE/Toronto: NT) with Alteon WebSystems (Nasdaq: ATON), and Sun Microsystems Inc. (Nasdaq: SUNW) are looking to add more integrated features to their systems as well.

What does it all mean for vendors of optical networking gear? Here’s my advice: Focus on the metro, with lots of capacity, 10-gig Ethernet ports, and intelligent interfaces between data and optical systems. This is happening already, driving the demand for metro DWDM from the likes of ONI Systems Inc. (Nasdaq: ONIS), and fueling interest in intelligent metro optical platforms from startups like Movaz Networks Inc. and Princeton Optical Systems Inc.. It also should build a fire beneath metro optical Ethernet, since these pure-packet solutions are ideal for connecting data centers natively over protected virtual Ethernet networks.

Within the data center, integrated systems will speed service creation and delivery, while creating demand for high-speed interfaces on core routers from Juniper Networks Inc. (Nasdaq: JNPR) and Cisco. The data center may actually be the first place to see a 40-gig interface on a router. In the virtuous cycle of optical networking, that means long-haul players with 40-gig interfaces on their DWDM (dense wavelength-division multiplexing) transport will have a lasting driver for their systems.

Virtuous cycles sounded great on paper over the past two years and fueled tremendous investment in optical networking gear, only to slow or stall in recent months. The reasons are complex enough for historians to argue over for years, but a simple explanation is the poor relationship between optical network deployments and revenue creation. In the present case, the coupling couldn’t be tighter: Data centers represent a real business opportunity for service providers, one that has already emerged as a strong Web-hosting business in the U.S. As storage, security, content distribution, and application hosting take hold, carriers can make a strong case for building out their metros at a pace that matches the rate of traffic growth. And that may just spark this market back to life.

— Scott Clavenna, Director of Research, Light Reading http://www.lightreading.com
khyron4eva 12/4/2012 | 8:27:32 PM
re: IDCs May Lighten the Gloom 1. Digex is never mentioned as a managed hosting
provider. Considering they are the premier
managed hosting provider, the only conceivable
reason for their absence here has to be that
they are now considered WorldCom/UUNet.

2. As well, even though they don't own physical
datacenters, LoudCloud is missing a mention. As
a former Digex employee and knowing some people
at LoudCloud, I would say that they should
be listed because LoudCloud's promise (which is
starting to be fulfilled) is that they are
"Digex done right". The biggest problem I
always had with Digex's model was that not
enough automation of processes was done.
LoudCloud's entire premise is to automate as
much of the provisioning, troubleshooting, and
management as possible. Their mistake was
bulking up as extensively as they did,
employee-wise.

3. OC-768 may first see the light of day in the
DC, possibly. But again, if the core can't
carry it, what's the point? How many carriers
are even ready to consider OC-768 at this
point? We couldn't get OC-192s in fast enough
at GlobalCenter (R.I.P.) when I was there. The
lead time was almost a year, if I remember
correctly. I mean, Qwest *just* announced a
full 192 backbone. Ask if those pipes are even
40% full. Doubt it. Even with companies like
Yahoo! and ftp.zdnet.com filling our pipes,
there was little need (and cost justification)
for 192 everywhere. That will change in time.

4. The ASP was a bad idea from the jump. Still
won't work. Storage is a different story. I
don't know about Corio, since they've never
appeared on too many radar screens I've seen
recently, but USi and NaviSite are endangered
species. And you sure don't see people jumping
to build new ASPs. Moving storage data across
those fibers has promise though, if done
securely.

5. All that DC space is not needed. Not yet.
Overbuilding will send some DC companies to
their deaths. A recent article in
the WashTech section of the Washington Post
talked about the huge amount of capacity
available in the Washington, DC area
that is going unfilled. I've seen similarly
here in SoCal. And I know that EXDS is just
waiting to be able to migrate as
many of GlobalCenter's customers out of those
GCTR datacenters into its own so it can start
boosting returns by shutting down facilities.

6. Managed services are the way to go. Genuity is
about to take some major Ls (Losses) as are
guys like EXDS and Intel, Above and IBM (didn't
they sell that business to ATT anyway - yet
another loser any way you slice it). You just
don't need that much. But the strong will get
stronger and the need and uses for unique
service creation abilities will return (if it
ever left). But I know I wouldn't trust my
stuff to EXDS and their managed services if
they can't keep a compromise from befalling
one of their own sites, and a security site at
that. Geez.
analyst-geocap 12/4/2012 | 8:23:40 PM
re: IDCs May Lighten the Gloom Khyron4eva,

What about the processes that DIGX has been speaking about on their last 3 conference calls. Do you think this is all smoke and mirrors, or is there some serious work being done? DIGX GM's and provisioning time have been moving in the right direction, these are indications that some effeciencies and scale are being driven into the model.

How long ago did you leave DIGX?

[email protected]
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE