The year 2007 saw Reliance Retail, one of the largest players in the Indian retail sector testing the usage of RFID (Radio Frequency Identification) technology in its processes.
While many hailed this as a big move in the growth of this already bludgeoning industry, the usage of such technologies still has many roadblocks and challenges.
Gemstone Systems, which has put forth the espousal of myriad enterprise infrastructure technologies has
Jags Ramanarayan, Global CIO of Gemstone Systems, speaking to
Prasad Ramasubramanian of
CyberMedia News on the prevailing turbulent economic conditions and the recent large deals bagged by his company.
Excerpts from the interview:
Could you elucidate on the challenges that are prevalent in the Stream Data Management and CEP (Complex Event Processing)?
One of the big challenges when dealing with unpredictable and high streaming event rates is scalability. Can the underlying infrastructure scale horizontally to handle bursts (1000's of events per sec) with a high degree of data consistency? Most systems cannot handle bursts and use techniques like sampling, potentially losing important data points in the analysis.
The other big challenge is the ability to look into history before acting on the incoming event. For instance, in e-commerce, the key to detecting suspicious activity in real-time lies in being able to inspect past access patterns from the same user. This means you need access to lots of data very quickly if the fraud detection software is expected to avoid significant impact to the user experience (response time). High event rates also imply that data can typically only be managed in-memory, raising concerns about continuous availability.
How does GemFire employ its expertise in providing a solution for CEP (Complex event processing) and stream data management?
GemFire's core strength lies in its ability to pool memory (and even disk) across a large cluster such that streaming data can be partitioned for data management as well as for complex event analysis. Built-in adaptive techniques constantly pay attention to the shifting load conditions and automatically re-balance data and behavior to meet throughput and latency targets in very high performance situations.
With a significant percentage of the historical event data kept in-memory, GemFire now enables applications to introspect contextual and historical data at main memory speeds. GemFire can ensure continuous availability of the applications by making sure the streaming data is managed in multiple nodes synchronously at all times. Failure conditions are automatically detected and clients failover transparently without any loss of events or latency degradation. GemFire offers continuous querying for filtering stream data, but its memory-based data fabric/Grid technology is quite complementary to most CEP engines.
With retail continuing to grow with renewed vigor despite the turbulence in global economies, how critical is it for players in the retail to adhere to RFID technology and how has GemStone helped its clients on the same?
Even though the pervasive use of sensor/RFID technology is still somewhat distant in the retail space, automated warehouse management is where RFID and CEP play an important role. The big value here for corporations is in avoiding exception conditions; wrong crate placed in the wrong rack, packaged shipped to incorrect address, etc. Even though low latency or very high performance isn't the real concern, scaling does remain a concern. The ability to handle events from hundreds of package handlers at any given facility and the correlation of that data across geographic boundaries are important features to preserve.
As the world devices newer technologies to propagate environmental friendly technologies, how is GemStone faring in this front? What are your initiatives?
GemStone's product strengths lie in optimal utilization of hardware resources like CPU and memory. By distributing data in-memory, GemStone's products like GemFire are able to significantly reduce the I/O wait for a transaction that in turn helps generate high CPU utilization. Our customers are seeing the benefits of high CPU utilization from our products through reduced wastage of CPU cycles and hence driving down energy consumption from their servers. GemStone has built a strong partnership with Intel where we are continuously testing our software against their newer generation of processors to provide greener solutions to our customers.
You have been one of the pioneers in the virtualization space and companies across India have begun embracing this technology like never before to spread their wings domestically, how has GemStone approached companies that are entering the virtualization arena today?
GemStone approaches these companies by promoting the vision of an enterprise wide data fabric. Most
CIOs today are faced with the challenge of introducing new technologies to remain competitive, and yet their overall IT budgets are being squeezed. Look around the Global 1000 landscape and it is amazing to see how much data is redundantly available where CIOs are forced to manage large replicated databases and other legacy repositories with significant maintenance costs. Virtualization of information through utility computing offers a significant advantage where the newer applications go through a fast, reliable data layer that sources, distributes and sinks data from a variety of data repositories.
Taking GemStone as an example, which has recently bagged enterprise deals with leading banks in the world, are we witnessing trends where the entities in banking space are streamlining their operations on a single vendor with a proven track record? Would other sectors follow suit too?
What we are witnessing is the recognition of financial markets that data fabric technology is an essential software infrastructure for managing data across front, middle and back office systems. What started several years ago as early experiments in this technology to address latency and performance issues in trading systems has now proven itself to be highly successful and has spread to encompass algo-trading, global risk management, market data distribution, reference data, real time pricing, global book and even use cases that we did not envision like application monitoring and HA/DR systems. Many banks are realizing the overall TCO value of sharing and distributing data in an "operational cloud" instead of traditional data management systems has allowed them to optimize hardware utilization, reduce database licensing, increase developer productivity and dramatically decrease the costs of database administration.( All this while realizing a 10X improvement in performance in real time systems). That is a "make money/reduce costs strategy" that under the pressure of the current economic realities is hard to resist.
Choosing a single, "pure play" vendor with a complete vision and focus on continuous improvement and enhancement of data fabric technology which has a proven track record in building and delivering enterprise systems is a pretty compelling value proposition to the banks right now. GemStone has hosted visits from many of the new prospects and partners from other sectors who want to talk to the Wall Street banks to learn about what they are doing. The clear message they are hearing from these top tier banks is accelerate your evaluation process and look across your entire enterprise for wide spread deployment opportunity of a data fabric. This is a game changing/must have technology for new real time systems development.
As a vendor, what were the changes in your clientele expectations, say five years ago and now?
The desire to have a real-time, global view of the data across multiple geographical locations is the single largest issue for our clients. This is the synthesis of OLTP (or Extreme Transaction Processing as described by Gartner) and OLAP/BI technologies. The ability of a data fabric like GemFire to combine the functionality of in-memory data distribution and near real-time event generation in one system along with support for virtually all data types has enabled it to meet the demanding needs of our clients.
As we are expanding our business, we are learning that other sectors wants to see the following characteristics in the product:
- Greater ease of use and insertion transparency of the product.
- More tools to support development and seamless integration with other software infrastructures.
- Seamless integration with other virtualization technologies to allow for a unified systems management and deployment operations.
- Availability of GemFire within sector-specific application software packages.
- These are the changes we see occurring in the next five years.
Cost cutting is the name of the game today, as a CIO how do you foresee the scenario with regards to IT spending in the quarters to come?
It is not a matter of cutting costs, but of optimizing the infrastructure currently deployed at FI's. If FI's focus their expenditures in the coming quarters on efficiency mandates and updating legacy system architectures, that will go a long way to cutting costs for the organization as a whole without stalling progress. Expenditures should be considered with an eye towards enterprise-wide coverage and not towards more costly silo efforts between departments.