NEWS

Telcos Reap Rewards of Xeround's Intelligent Data Grid

On-Demand Enterprise
September 5, 2008
By Derrick Harris

Stop me if you've heard these terms before: data grid … database virtualization … data management … you get the point. Well, it turns out there are different ways to tackle the same problems, and the telecommunications space currently is reaping the benefits of one of the newest attempts to solve the distributed data nightmare.

The solution du jour is Xeround's Intelligent Data Grid, a product that goes beyond hardware-footprint constraints of traditional data grids and actually allows for a federated view of data scattered across the distributed network. The result, says Xeround CEO Charlotte Yarkoni, is more of a "database virtualization engine" that spans multiple existing data stores and creates on logical view of the subsets of data. The company currently is targeting the "communication service provider sector" -- that is, anyone with subscriber data that needs to be managed -- with the biggest fans thus far being the telcos, who often have subscriber data spread around globe and no real way to make sense of it in real time.

According to Ari Banerjee, director Yankee Group's Enabling Technologies Service Provider division, more than 80 percent of telcos believe subscriber data management is a big problem, mainly because better subscriber data management means more revenue. Maximizing subscriber data enables them to provide targeted services and advertising, location-based services and other alternative business models. Wireless providers, for example, could provide a Red Sox fan with pertinent information, coupons, etc., that could be of use to that subscriber while traveling outside Boston. However, says Banerjee, the window of time to provide such services is small, and having data stored in separate silos inaccessible to one another makes rolling out this new business model difficult. In a recent Yankee Group survey of 150 Tier1 and Tier 2 service providers, more than 60 percent of respondents cited either "inability to personalize the experience" or "inability to have a common view of subscriber background" as the biggest factors inhibiting innovation.

Another example Banerjee points to is the customer service experience, where finding just one phone number to handle all of your issues is a rarity. "I buy cell phone service and landline service from, say, AT&T," he hypothesizes. "You have to have two [customer service] numbers to call if you want to complain, just because they don't have visibility; it goes back into that siloed environment. … That also means that when you call and you talk to them as a wireless customer, they don't know that you also are an existing wire line customer because the data of your wire line usage, etc., is stored in a different domain."

The traditional methods for solving this problem have been either centralization of subscriber data or, unfortunately, doing nothing at all. Banerjee says less than many but "quite a few" telcos have invested in centralization, but even though it might be the promised land, many stay away because they view the process as expensive, time-consuming and potentially disruptive to business. Centralizing all of a provider's data stores is "like open heart surgery," Banerjee says, and the process can take months to more than a year. On top of that, he added, unless all of a provider's legacy systems and platforms are retired (which is rarely to never the case), subscriber data still will be spread across the network, even after a multi-million-dollar transformation project.  Replication also has been attempted, but it is a cumbersome and not wholly effective process. "That's why we see the mentality of service providers," he explains. "Typically, they're more reactive than proactive, and it's always been that way."

The other approach, says Banerjee, is data federation, which is "shorter-term, less expensive and … faster to implement." It also is less disruptive and allows service providers to centralize their data gradually instead of all at once.

And in steps Xeround …

Four Key Differentiators

Xeround's Yarkoni says the Bellevue, Wash.-based company's Intelligent Data Grid (IDG) distinguishes itself from other data management solutions in four key ways: heterogeneity, grid computing, geo-distribution and self-healing.

1. Regarding the first of these, Yarkoni says IDG was designed with heterogeneity in mind across the board, from different datacenters to different machines to altogether different types of data. "First and foremost, what we recognized [was] the world is getting more heterogeneous, and it would continue to get heterogeneous," she explains. "This is very important in terms of, first, just accessing data from a variety of these elements."

In the past, she says, telco networks were pretty straightforward, but more elements were added as the number of offered services grew, with each element sitting in its own network location with its own customer data. Not only that, but, increasingly, these disparate elements don't even house the same type of data. Some are relational databases (SQL), some are hierarchical databases (LDAP) and "in more recent times," says Yarkoni, "we've seen a lot of the content platforms that get produced use XML as the way they store their data, which is a completely different format."

"The analogy you should think about," she explains, "is: You're trying to build a set of database capabilities, but you have to be able to speak English, Russian and French simultaneously to look at that data simultaneously, because each one of them are stored in a variety of formats."

IDG addresses this problem with its N-way Access Protocol, which lends the ability for the software to interact natively with different data formats. Yarkoni acknowledges that other products can interact with multiple formats, but that usually has requires some type of add-on tool, which can have an impact of performance and might not address future data formats. "We don't have to have a translator sitting there with us to have a dialogue with each one of these entities [SQL, LDAP, XML, etc.]," she says.

2. By applying its grid computing algorithm to the database layer rather than the physical layer, a process Xeround calls "virtual partitioning," Yarkoni says the company has created a powerful product that expands on the traditional data grid notion of having to keep that data within a defined set of physical resources. "At a very basic level, when you have a data store, a physical disk that sits out there, that disk comes with an index," she says. "If you want to do any type of I/O with that disk, you route through that index to understand what ones and zeroes sit on that disk. This is the same concept that we apply to our software but, again, we did it at a database level so that we have the ability to look at multiple physical data stores -- again, regardless of their format -- and because we have one virtual index that accesses them, we can treat them all as one logical entity on the backend."

3. Xeround's patented geo-distribution technology enables IDG to "function with LAN latency over [a WAN] at any given time," says Yarkoni. Telecommunications companies take fault tolerance and redundancy very seriously, so they always have multiple copies of data on the network. IDG exploits this redundancy by finding the shortest route on the network to the closest copy of the data. While she admits this is similar to data caching, the difference, she says, is that many caching solutions are read-only, whereas IDG allows for true read-write I/O and synchronization -- a two-way street, if you will.

4. The self-healing properties of IDG should be apparent. Because it is not tied to any specified hardware, it is not affected by any particular box going down or going offline. In addition, Yarkoni adds, this means IDG is linearly scalable, and no reconfiguration is necessary if more hardware is added to the mix.