Migrating from a legacy mainframe environment to a distributed environment brings new freedom and integration while enabling the adoption of microservices and easing digital transformation. These are all great things. However, modernization is not without its challenges. For example, with a distributed systems architecture, databases and applications reside on separate servers. In a legacy mainframe environment, they can exist in tandem. The separation of these entities can introduce a small amount of latency when comparing the new to the old. In large batch processes, this latency can significantly affect performance.
Situations such as the introduction of latency through data access are universal, they come with the territory, and without the experience and expertise of a professional modernization team, remediation can be incredibly difficult. Here at Modern Systems, we have run into and resolved varieties of this issue a number of times, most recently with a large American media company.
Post-migration, one of their core batch jobs was taking far too long to process. The Modern Systems team set to work and customized a mechanism of caching so that data is loaded in parallel in the cache, which significantly reducing network round-trips and, thereby, greatly reducing the aforementioned latency. This VSAM in-memory cache solution enabled caching for batch of the specified files (enabled in the job xml as a step specifying file names, or using dependency injection). This customized cache layer has the following features:
- Parallel update on separate update thread
- Low memory footprint – data will be stored as bytes, each index will have only a sorted reference array
- Fast access of key data – using binary search and byte comparators for positioning, only “in-memory”
- Same transactional model as the batch (after step commit)
- Support for both multiple record layout types
- Data access if relevant data is loaded, even before the cache is completed
Once implemented, the supercharged batch caching solution reduced the job processing time by 90%! What’s more, nothing in the customer’s application code had to change. All of the customization took place in the data access layer, retaining the 100% like-for-like nature of the application code conversion we’re so famous for.
Employing modernization expertise is a crucial part of reducing scope, time, and risk associated with retiring big iron. Our team at Modern Systems is a trusted advisor to all of our customers along the way to ensure a smooth transition. In fact, we have completed over 200 successful modernizations over our 30 years in business.
Interested in learning more about why our people are the key to our success? Contact us.