Programmers: Pervasive’s Parallelization Provides Punch, Profit

After 27 years of steady growth, Austin, Texas-based Pervasive (PVSW) has become a $47M annual run rate software provider. Its portfolio includes a “zero admin, light footprint database” (the former BTrieve, now PervasiveSQL), data integration software (for SaaS and on premises applications), and data synchronization products for such apps as, Quickbooks and Microsoft Dynamics CRM. In 2009, it began leveraging its DataRush processing engine as a product, providing a solution for companies that want to take advantage of multicore architectures to drive dramatically enhanced performance on much smaller footprints, for programming data services tasks such as aggregation, de-duplication, cleansing, integration, matching and sorting, as well as data mining and predictive analytics.

Pervasive has racked up 36 consecutive profitable quarters over the decade, and through the last couple of years has seen a significant ramp up in revenue. Claiming over 4500 customers and over 1000 SaaS implementations, it projects a sense of being in the right place as the market catches up to it. It’s moving its portfolio steadily into the cloud – the Pervasive DataCloud, running on Amazon today. But it remains little-known to the general IT community. The database side of the firm represents 2/3 of its revenue and has been steady, but its database brand is rarely visible.

Why the stealth model? “We do sell our data products to large corporations, as well as through SIs and SaaS vendors, but we designed them to be embedded,” Jim Falgout, Pervasive DataRush chief technologist, told me in a recent briefing. “We can compete with majors like Informatica, IBM and Oracle on product, but that’s not our primary focus. The database, for example, is used almost entirely for embedded applications – that’s a different market and a different selling model.” Today, although the other Pervasive product teams emphasize brand, the majority of Pervasive’s sales are through the channel, and growing faster there than through direct sales. Over 150 Pervasive Integration ISVs including McKesson, Sage, Epicor, Intuit, Metavante, Eloqua, Daptiv, and Xactly are using the firm’s software. System Integrators such as Ceridian and CSC in vertical markets like healthcare and telecom are also an area of focus. Pervasive focuses aggressively on making the connection to its partners when inquiries come in, both in North America and internationally.

The flagship Data Integrator offering (which began many years ago with the acquisition of Data Junction) provides connections to a huge number of data sources. Pervasive’s ISVs and SaaS vendors put it through its paces for connections to databases and applications for B2B integration using a variety of industry standards including  HL7, HIPAA, EDI, NCPDP, FIX, UCCnet, ACORD, as well as various XML and flat formats.  In the past few years Pervasive has been increasingly used for connecting SaaS applications at the periphery, as within the SaaS stack. The company asserts that it has been a partner of longer, and has more in-production customers, than any other vendor offering salesforce integration.  DataCloud offers the ability to do specific integration and synchronization jobs without any on-premise software. It also includes pre-processing capabilities, such as DataProfiler, from Pervasive’s broader portfolio of data integration products.

The new news is DataRush, and the equation is simple and persuasive: more cores = more speed. And no rewrites needed when moving to similar processor architectures – even across different operating systems. For programming organizations that are using R for statistical projects, or coding ETL and data analysis work with MapReduce, the performance benefits claimed are very attractive. Nena Marín, Ph.D. Pervasive DataRush chief scientist, described the software platform as java-based, relying on a message passing architecture. Difficult problems like deadlock detection, threading, and resource management can be roadblocks that prevent programmers from getting the full value of parallel architectures.  DataRush makes it easier to code than using Java primitives. Pervasive also provides profiling and debugging tools, tackling an often daunting challenge for many-threaded app architectures.

Our conversation veered into discussions of AMD’s Code Analyst, which allows programmers to take info from the JVM and combine it with processor information to look at such useful indicators as cache misses and overflows. By this time, I was deeper into the internals than I wanted to be, so we returned to a discussion of use cases – and things became even more interesting. In today’s market, where running code against files is often a viable alternative to building a permanent data warehouse, Pervasive’s performance benefits offer sizable advantages, and the claims are compelling. Parallelizing specific algorithms outperformed R and MapReduce jobs in dramatic fashion in some cases Pervasive showed me.

A note of caution: at this stage, Pervasive has few attributable customer stories to show, so much of the material presented was internal tests and benchmarks. I’m told Pervasive has now completed implementations for some unnamed customers (and have others in process) using Pervasive DataMatcher, a solution built on Pervasive DataRush to provide fuzzy matching on large datasets with rapid throughput.  The usual caveats I offer with no named references apply here: nothing substitutes for your own data and your own business problems. Pervasive offers a free download; get your hands on it and run your own tests.

DataRush is a new business for Pervasive, and is being treated like one: the team is relatively standalone. The division has its own sales and field organizations and its own R&D. Pervasive is clearly not yet funding dramatic growth; it seems content to expect DataRush to be “significant within 2-3 years,” recognizing the long sales cycle in its channel and ISV model. This is likely to be a self-fulfilling prophecy unless Pervasive makes significant additions to its current marketing. A few case studies, some focus on reaching the programmer community through social media, events and publications could go a long way. It’s unlikely that Pervasive DataRush will remain the only provider of such services; it needs to capitalize on its lead now and gain a branded foothold.

Disclosures: Pervasive Software is not a client.

Published by Merv Adrian

Independent information technology market analyst and consultant, 40 years of industry experience, covering software in and around the data management space.

One thought on “Programmers: Pervasive’s Parallelization Provides Punch, Profit

Leave a Reply

%d bloggers like this: