Cellular In The Enterprise Data Center

As too many enterprises have learned the hard (i.e., expensive) way, when an old-school modem attached to remote data center equipment gets unplugged, far-flung network infrastructure issues can suddenly become decidedly hands-on issues. And last-minute plane tickets to out-of-the-way locales aren't cheap.

Even beyond the disaster scenarios, POTS for enterprise data centers is surprisingly costly to maintain, regardless of whether or not there's a complication. As the prices of cellular data plans throughout the world continue to drop, the comparative cost of installing an additional T1 or T3 line solely for the purpose of network failover becomes more and more difficult to justify. T1 lines are dedicated services with ties to particular providers with long contracts attached, whereas with cellular, it's much simpler to change services or carriers.

Add cellular's advantage of availability -- with services able to reach into a data center wirelessly and fix network failures -- and the expected cost to repair a data center issue begins to dwindle.

Cellular also offers simplicity for data centers. Cellular out-of-band is independent of the production and management of fixed, wired networks, making possible consistent deployments across enterprise locations, even with differing network gear or available 3G/4G carriers.

"Cellular is becoming increasingly reliable and affordable. In fact, some regions are eliminating telephone lines or skipping telephone installations altogether and installing cellular, especially for remote management administration," says Mary Galbavy, the global operations director for USRobotics. "At the same time, IT infrastructure is mission-critical, and downtime is unacceptable, so reliable out-of-band connections must be in place, even with limited IT resources and budgets."

Though cellular-enabled datacenters are certainly the future, they have their own growing pains to overcome, and there are scenarios where POTS lines can be valued as the proverbial turtle in the race: 56k slow but steady. Cellular coverage that varies across different distributed network sites can require solutions using multiple carriers to connect. In addition, the quality and speed of the fastest service, 4G LTE, is not guaranteed for every cellular connection through a carrier, meaning it is not a reliable sure thing.

International LTE rollouts can be further complicated by differences in the technology used by different carriers, though those gaps are always shrinking. These discrepancies can produce experiences of 4G LTE with 3G-HSPA+ or 3G-EVDO fallback, and users may hop between 4G and 3G connections to a potentially frustrating effect. The amount of time a network admin remains on 4G versus falling back to 3G can vary a great deal and can account for LTE speeds lower than expected.

Meanwhile, cellular providers are acting upon incentives to move customers off POTS lines and on to cellular networks. At first glance, this may seem counterintuitive, since it amounts to moving customers to less expensive plans, with providers favoring cellular in circumstances where they might be migrating users from $60/month to $5/month billing.

However, providers with customers using landline services must pay the incumbent local exchange carrier (ILEC) for the use of each line while having no control over the related billing or quality. This situation further drives telcos to encourage adoption of cellular infrastructure management solutions. Support for cellular will continue to get better; support for landlines will not.

At the same time, recent moves by cellular providers, though made with a bright future in mind, may contribute to uncertainty in the present. Verizon plans to drop CDMA/EVDO, but that seems impractical for critical network failover or out-of-band management in the short term. AT&T is refarming 2G/3G bands to convert their use to LTE, but it is not abandoning 3G entirely. The goal is an eventual rollout of guaranteed LTE internationally.

But a stumbling block even larger than the differences in carrier technologies may be the expansive list of LTE spectrum bands in use by carriers in different countries (800 MHz, 900 MHz, 1,700 MHz, 1,800 MHz, 1,900 MHz, 2,100 MHz, 2,300 MHz, and 2,600 MHz). Efforts to civilize this mess are in the works, with most countries outside the US looking to standardize on the 700 MHz (APT band 28) spectrum and including a two-duplexer system in devices and network equipment. If this goal were realized, it would take the uncertainty out of LTE.

An advantage of LTE is its efficiency in using the radio spectrum, compared to 3G or previous technologies. LTE uses less power, has scalable bandwidth, and is all-IP, which reduces latency, with a transition of less than 100 ms from dormant to active. LTE is IPv6 and IPv4 capable, so it is ready to handle the explosive growth of Internet-capable devices.

The LTE infrastructure itself includes authenticated and encrypted internal signaling with integrity protection, along with enterprise end-point routers and appliances with built-in firewalls, strong authentication, and VPN tunneling with encryption for "over the Internet" connectivity. Data centers will be drawn to data plans featuring support for non-Internet-facing private 4G LTE carrier networks. Data center customers will enjoy the carriers' guarantee that their traffic will be segregated from the Internet if they use a private network plan and 4G.

LTE faces issues like any evolving carrier service, but most 4G-enabled products already cover AT&T, Verizon, and Tier 1 international LTE bands plus ubiquitous 3G and 2G fallback. Eventually, we hope the final LTE bands will operate the same in any country, because that reliability will directly contribute to more reliable remote data center management.