Data Center Knowledge | News and analysis for the data center industry

Data Center Knowledge | News and analysis for the data center industry - Industr's Journal

[Most Recent Entries] [Calendar View]

Tuesday, May 24th, 2016

Time Event

1:00p

Ahead of IPO, Nutanix Makes Hyperconverged Play for SMB Market

If last December’s filing by hyperconverged platform maker Nutanix for an initial public stock offering was supposed to have been the start of a drum-roll, the poor fellow whose arms are about to drop through his timpani got some good news on Tuesday: At Citrix’ Synergy conference in Las Vegas, Nutanix announced a more bite-sized version of its cloud platform.

It’s a market play directed not toward its usual Fortune 2000 customer base, for whom the “hyperconvergence” watchword doesn’t trigger the spell checker on the word processor. Rather, it’s an introduction of the Nutanix brand to small and medium businesses — those who only run a few VM instances at a time, and who today are being served by Amazon and its public cloud competitors.

It’s the type of move that could round out Nutanix in the eyes of potential investors, some of whom may be skeptical of whether a company so focused on hyperconverged appliances may end up being a one-hit wonder.

“What we’re launching is essentially all the Nutanix goodness, but at a better price point that would make it more preferable for small and medium businesses,” said Prabu Rambadran, Nutanix’s director of product marketing, discussing his company’s new Xpress service in an interview with Data Center Knowledge.

Converging Something Else

Xpress, Rambadran said, will include compute, storage, virtualization, systems, and operations management in a three-year support package with a base price of $25,000. The service is designed, like its older and bigger siblings, to be hypervisor agnostic, supporting VMware ESXi and Microsoft Hyper-V, as well as the company’s own Acropolis hypervisor (AHV).

Other vendors in the enterprise space, he remarked, have strategies similar to the way a big truck manufacturer makes a play for the consumer space: by overloading the chassis with air conditioning, sound systems, and in-car electronics. He implied that they’re adding bullet points to their product SKUs without adding value that would actually apply to customers’ everyday workloads.

“What we’ve done is, effectively, not that,” said Rambadran. “If you think about the core storage and virtualization capabilities that SMB customers need, and compare them with what large enterprises need, it’s very similar. All of them need simplicity of management; all of them need high levels of availability, four- or five-nines; and all of them need a solution that shifts the focus back to applications.”

Large enterprises have had the advantage of being able to assign one IT administrator for specific elements of the operation, like a specific application, or a dedicated bank of servers, or even just printing, he pointed out. But that’s becoming less and less of an advantage anyway, as major data centers like Google’s, Amazon’s, and Azure’s have demonstrated the viability of one admin for every 500 or 1,000 servers. These major cloud providers have deployed self-service portals, he pointed out, that have reduced the need for service delegation in the IT department, in a way that applies just as much to enterprises as to SMBs.

Distinguished Workloads

Nutanix’ new marketing push is this: Maybe SMB and Fortune 2000 customers have different needs today, and the phrase “data center” typically applies only to the latter. But as this broader convergence trend plays out, the services they’ll require will end up becoming homogenous slices of power, served in different quantities.

This may indeed be how this market plays out, assuming a VM is a VM is a VM. However, as some analysts have already noted, the nature of workloads that are virtualized by VMs can be very different between enterprises and SMBs. Specifically, enterprise-class virtual machines are bound by enterprise policies that directly address how they interoperate with other VMs, with hypervisors, and with underlying hardware for security reasons.

“Consider the state of centralized policy in the data center,” wrote Hyperconverged.org analyst Scott Lowe, for his 2015 State of Hyperconverged Infrastructure Market Report [PDF, registration required]. “For data centers that have equipment from a wide variety of vendors or that have a lot of ‘point solutions,’ such as WAN accelerators and replication tools, there could be a number of touch points when it comes to policies. These various touch points don’t always align very well with one another, particularly when there are different vendors in the mix. For example, while it may be possible to define some policies at the hypervisor layer, it’s often difficult to apply storage policies that have any awareness of virtual machine boundaries.”

VMs for data centers include not just workloads but virtualized appliances, actually forming some of the infrastructure upon which those workloads are hosted. Managing data center-scale virtualized systems will certainly require a different kind of administration — not just the same admin who can finally become two once the threshold of the 1:500 or 1:1000 admin-to-server ratio is finally reached.

All this doesn’t even touch on the whole discussion of containerization, which Nutanix’s Rambadran believes only applies right now to the cutting edge of large enterprise data centers. Working in Nutanix’s favor for the moment may be another fact that he acknowledges: Most of these big data centers have not yet pushed containerized infrastructure into production. “The jury’s still out on that,” he said.

“One of the reasons why small and medium businesses typically look at a public cloud solution as a viable alternative,” he continued, “is because standing up an infrastructure in their data centers has always been complex [and] expensive.”

SMBs, he believes, won’t consider containerized solutions just yet if they’re looking to deploy, say, an application based around a relational database, because they may not address this complexity factor. Rather, they’ll look towards moving their VMs into an infrastructure that’s more cost-effective and simpler to manage. So while there’s a strong argument that containers could provide a cost-effective infrastructure solution for database workloads in the long term, the convergence of enterprise goals and SMB goals may take place in the shorter term, around deploying first-generation VMs more competitively against options like AWS.

That’s a different message from the one Nutanix has given to its traditional enterprise customers, one which touts the “intersection” of Docker containerization and DevOps. What Nutanix appears to need right now, in order to fight the ongoing perception of declining IPO value overall, is some public evidence of a kind of convergence that investors, not DevOps engineers, will appreciate.

Tuesday’s announcement was coupled with the news, delivered at the Citrix Synergy conference, that Nutanix will begin offering a turnkey solution for deploying Citrix virtual desktops, called InstantOn. Geared toward mid-sized organizations, the service will be offered at a flat base fee of $415 per desktop.

(Comment on this)

3:00p

Metro Optical Networks at the Heart of Internet Exchanges

Jay Gill is the Principal Product Marketing Manager at Infinera.

Internet exchange (IX) is the heart of the Internet, the place where the Internet actually gets connected. Increasingly, IX facilities are also the focal points for content delivery, cloud-based service delivery and enterprise hybrid cloud architectures, attracting new customers and driving new interconnection models.

This rapid change in the Internet exchange ecosystem is driving explosive growth in bandwidth between IX sites as well as the demand for massively scalable metro optical networks.

When we look inside the metro IX ecosystem, there are four major drivers for metro optical bandwidth:

Metro IX interconnection
Extending the cloud edge
Point of presence (PoP) economics
High-performance hybrid clouds

Metro IX Interconnection

The largest Internet exchange players have multiple locations in each major market, either in facilities they own or in leased space of other facilities. For example, Equinix and Interxion use a mix of owned and leased facilities. Twenty-five of the 33 markets served by Equinix have more than one Equinix IBX facility, while Interxion operates multiple facilities in five of its 13 markets. Other providers who focus solely on the business of Internet exchange may locate exclusively in leased space, but they too may have multiple locations in a market. For example, German-based DE-CIX has 20 “DE-CIX-enabled” locations in Frankfurt, most of them located in facilities of other IX providers, including Equinix and Interxion.

Internet Exchange Provider	Number of Metro Markets	Markets with >1 Location	Examples (Number of locations)
Equinix	33	25	Washington, DC (10)</p> New York (8) London (6) Frankfurt (5) Tokyo (4) Hong Kong (3)
Interxion	13	5	Frankfurt (9) Paris (7)
DE-CIX	9	5	Frankfurt (20) New York (8)

All of these IX providers interconnect their sites with high capacity metro optical links, turning the multi-facility footprint into one big distributed exchange and offering their customers more potential partner connections.

Extending the Cloud Edge

An increasingly large majority of traffic delivered to Internet users comes from a small number of hyperscale cloud and content providers. In North America, Sandvine reported that around two-thirds of the traffic came from just seven providers, with Netflix leading the way, accompanied by well-known players such as Facebook, Google and Amazon. All of these providers are expanding their cloud networks to reach more and more major metro locations where they can connect directly to the access networks that serve their end-users.

Many are building or leasing data center infrastructure and they often need multiple data centers in a single metro area to get around space and power limitations. They also need connectivity into multiple IX facilities to hand off traffic to access providers. For the best customer experience and for operational efficiency, they need all of their data centers and IX locations within a metro to operate as if they were a single, distributed data center, and that requires very high interconnect capacity.

Some of the metro interconnect links deployed by the hyperscale cloud and content providers are the largest found anywhere in the world, scaling to tens of terabits per second and driving demand for the most advanced optical interconnect technologies available. To address this demand, Ovum and ACG predict that the compact, high-capacity data center interconnect equipment market will grow at an average rate of over 65 percent per year for the next three to four years.

PoP Economics

Given the popularity of Internet exchange sites, the costs of colocating in them are often high relative to other data center and colocation options. This drives a wide variety of IX users to locate their primary infrastructure in cheaper data centers near the IX facilities and then link to a small interconnection point of presence within the IX. A good example of this strategy is IBM’s Softlayer division, which provides a range of managed hosting and private cloud services from its own data centers and promotes its high capacity optical interconnects to popular IX sites as a key part of their total solution.

High-performance Hybrid Clouds

A fast growing use case for private metro optical networks is the hybrid cloud. As noted above, private cloud providers are connecting to IX sites at high capacity to enable rich Internet connectivity, but they are also increasingly using that capacity to provide direct connectivity to the public cloud providers, such as Amazon Web Services, Microsoft Azure, and Google Cloud. By doing so, they can offer their enterprise customers a flexible mix of private and public cloud services with the ability to easily expand capacity and migrate applications between clouds.

Some enterprises are taking matters into their own hands, leasing space in IX facilities to deploy their own data center infrastructure with direct access to their cloud service providers, and connecting that back to their enterprise data centers using high capacity optical networks. Equinix has said that such enterprise customers will be their fastest growing customer segment over the next few years.

Whether enterprises pursue an outsourced model to build their hybrid clouds or do so in-house, the key requirement is direct connectivity to their cloud providers with high performance, low latency and inherent security, and metro optical networks are a great fit for these requirements.

In summary, several major trends are driving demand for Internet exchange services as well as the need for high-capacity metro optical data center interconnection into and between IX facilities. It’s time to recognize that the combination of IX plus metro optical data center interconnect is how today’s Internet gets connected.

Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.

(Comment on this)

6:02p

Data Center Market Spotlight: Northern Virginia

Our theme this month is site selection. From electricity costs and network infrastructure to the available pool of skilled workforce, data center site selection is one of the most complicated and important business decisions a company makes. Data center location affects everything from the cost of doing business and overall company agility to the quality of user experience. And, like every other aspect of the data center business, where companies choose to put their critical IT infrastructure and why is changing because of … you guessed it: the Cloud. This month, we’ll examine these trends more closely.

Of all North American data center markets, Northern Virginia has been the most consistently active one, due in large part to its history. In the early 1990s, the region played a crucial role in the development of the internet infrastructure as we know it, naturally drawing a high concentration of data center operators who could connect to many networks in one place.

Today, fueled by the big cloud providers’ race to expand data center capacity, the market is hotter than ever. To say that demand for data center capacity in Northern Virginia is outpacing supply would be downplaying the situation.

“Northern Virginia is as tight as I’ve ever seen it,” Jim Kerrigan, managing principal at North American Data Centers, says. “If you want space right now, you’re in bad shape.”

NADC is a commercial real estate company specializing in data center space, and Kerrigan helps companies, including some of the biggest cloud providers, source data centers.

The reality is that wholesale data center developers in Northern Virginia cannot build facilities quickly enough to address the demand. It’s typical for a cloud company to take down 10MW or 20MW at a time, quickly leaving the market in short supply. Less than 10MW is currently available in the area, according to Kerrigan.

In the first quarter alone, Microsoft leased 22MW of capacity in Ashburn, which is where most of the region’s data centers are, according to a recent market report by NADC. Six of last year’s 20 largest wholesale data center leases signed in North America were in Ashburn. The lessees were Microsoft, Apple, Oracle, Uber, Google, and PNC.

Demand in Northern Virginia in 2015 totaled more than 60MW of data center capacity, according to a January 2016 report by Jones Lang LaSalle, a commercial real estate firm with an extensive data center practice. For comparison, companies leased 42MW in wholesale capacity total in the Dallas-Fort Worth market last year and 32.5MW in the Chicago metro.

The Construction Race

Because there’s more demand in Northern Virginia than anywhere else in the US, there’s also more data center capacity under construction than in any other market. Close to 30MW was under construction in the region as of the first quarter of this year, according to Digital Realty Trust’s first-quarter earnings presentation. Digital is one of the biggest players in the Northern Virginia market.

But, because capacity is generally taken down as soon as it comes online, there’s little worry of oversupply. “Current construction pipelines are generally concentrated in top-tier national markets with high visibility on demand, and pre-leasing levels are healthy,” William Stein, Digital’s CEO, said on the company’s first-quarter earnings call in late April. “In addition, market vacancy rates within these markets are in the single digits, and our own portfolios in these same markets are likewise north of 90 percent leased.”

In other words, the providers’ existing capacity is maxing out, and they’re building because they see clearly that the demand is there.

Sabey Dips Toe in the Water

One of the companies building in Northern Virginia is Seattle-based Sabey Data Centers, which held a groundbreaking ceremony in Ashburn Tuesday. This is going to be Sabey’s first data center in Northern Virginia, and the company is threading lightly, starting only with 2.5MW of turn-key space.

The region is “the best market in the world, by pretty much any measure,” Robert Rockwood, Sabey’s eastern region head, says. “We see an opportunity to bring Sabey-quality space into the market now, at a time when absorption is outstripping new construction to a meaningful degree.”

Acknowledging that 2.5MW is a “small first bet,” Rockwood says he feels it is the right move for a private company coming into a market without having customers already signed. Sabey does have existing customers in its other markets, in Washington State and New York City, who are interested in taking space with the provider in Northern Virginia.

Also, the initial 2.5MW of turnkey space that’s expected to come online by the end of the year will be the first data hall in a four-hall building shell. Once the building is outfitted with infrastructure, it won’t take as long to complete the remaining data halls using Sabey’s modular design, he says.

DBT-DATA Buys Land for Construction

One of the developers in Northern Virginia that has kept a low profile but scored some big wins in recent years is DBT-DATA. Two of the company’s operational data center properties in the market are leased to RagingWire, the NTT Communications-backed data center provider, and this week, the developer announced acquisition of two additional properties in the region where it plans to build quickly.

“There’s a race among cloud providers to create capacity,” David Tolson, DBT-DATA’s CEO, says. The cloud companies’ business model has been validated, switching on the green light to make the big capital investment necessary to support their growth.

“If you’re going to make that incredibly large capital investment, why not go to the largest, most proven marketplace in the United States,” he says, referring to Northern Virginia.

Pricing Remains Low Despite High Demand

Interestingly, data center pricing in Northern Virginia has remained relatively low, despite the supply-demand imbalance. Some of the recent wholesale deals in the region were in the $105-$115 per kW range – some of the lowest monthly rates in the country, according to NADC.

Kerrigan says there are probably multiple reasons for this. One is that new players have been entering the market and using low rates to compensate for the newcomer’s disadvantage.

Another potential reason is the size of the deals the big cloud providers make. A 10MW deal with a top internet giant is very attractive, and even established players are willing to compromise on the per-kW rate if it means closing a deal like that.

Finally, some of the providers in the market offer a mix of wholesale and retail colo services, and they can afford to lease wholesale at lower rates, compensating for it with income from retail, which yields higher revenue per square foot.

(Comment on this)

7:33p

Report: Data Center Outage Cripples Virginia DMV

The State of Virginia Department of Motor Vehicles lost access to its IT systems for hours this past Saturday as a result of a data center outage that disrupted network access for more than 60 state agencies.

Caused by a faulty network switch, the outage was resolved about five hours after it started, Richmond Times-Dispatch reported. The data center is owned and operated by Northrop Grumman, which resolved the issue by replacing a faulty part with one from a test environment that was running in the facility.

The switch failure “caused a widespread outage of inbound and outbound communications traffic through the information technology infrastructure for executive branch agencies,” Marcella Williamson, spokesperson for the state’s IT agency, told the Times-Dispatch.

The outage is likely to further corrode an already sour relationship between the agency and the contractor. The agency, VITA, is in the process of getting out of a 13-year, $2.3 billion services contract with Northrop Grumman, signed in 2006, according to the report.

It wants to switch to a more distributed approach, using multiple service providers, and Northrop Grumman has declined to play part in that new arrangement, even as lead contractor.

As of Monday afternoon, the IT agency was unclear about the cause of the switch part’s failure. It also hadn’t established why the data center’s backup system didn’t work as designed.

(Comment on this)

8:19p

Inside the Facebook Data Center in Sweden (Video)

Few companies design and build data centers at Facebook’s scale, and the social network’s three-year-old computing facility in Luleå, Sweden, is one of the biggest in the world. As Bloomberg’s Ashlee Vance put it in the video below, you need a bicycle and a picnic lunch to tour the site.

The video, where Vance visits the Facebook data center in Sweden, accompanied by Joel Kjellgren, who has managed the site since its launch in 2013, is part of a series Bloomberg recently produced on the country’s booming tech sector, which has created hits like Spotify (one of the videos features an interview with the music streaming service’s founder Daniel Ek), SoundCloud, and King (makers of Candy Crush).

Check out the video below, as well as the vast DCK archive of Facebook data center coverage, including the Luleå site.

HPE’s Profit

Separately, HPE said fiscal second-quarter profit excluding certain items was 42 cents a share, on revenue of $12.7 billion. Analysts on average had projected profit of 42 cents on sales of $12.3 billion, according to data compiled by Bloomberg. In the current period, which ends in July, profit will be 42 cents to 46 cents, the Palo Alto, California-based company said. That compares with an average estimate of 48 cents.

Revenue in the Enterprise Group, HPE’s largest unit, posted a 7 percent gain from a year earlier to $7 billion. Sales in the Enterprise Services division being combined with CSC fell 2 percent in the April quarter to $4.7 billion.

(Comment on this)

<< Previous Day 2016/05/24
[Calendar] Next Day >>