Terracotta In-Memory Upgrades Add Speed, Capacity, Control for On-Premise, Cloud Apps 
Exploding data volumes are driving the need for better use and automation of in-memory resources, say the caching experts at Terracotta. The firm’s latest in-memory software aims to improve response times, scalability and service levels for on-premise and cloud-based apps. IDN explores Terracotta 3.6’s BigMemory and Automatic Resource Control with execs.
Exploding data volumes are driving the need for better use and automation of in-memory resources, say the caching experts at Terracotta. The firm’s latest in-memory software aims to improve response times, scalability and service levels for on-premise and cloud-based apps.
IDN explores Terracotta 3.6’s BigMemory and Automatic Resource Control with execs.
“Traditional application architectures are stressed by exploding data growth and new mobile and cloud projects, which can move data depending applications to the cloud even as the data itself stays local,” Mike Allen, Terracotta’s head of product marketing told IDN. “Our new offerings will give applications better access to more in-memory data, as well as provide IT automated tools to manage it.”
At the heart of Terracotta 3.6 are two key complementary features:
- An upgrade to its BigMemory technology, which bypasses Java garbage collection to enable individual caches over 1 TB per JVM; and allowing a single Java application to access tens of TBs at micro-second speeds;
- A new Automatic Resource Control (ARC), which provides valuable size-based cache control and automated management features to make it simpler and easier to tap in-memory resources. As a result, devs or admins can simply specify how much memory they want to user per application, and ARC implements those instructions.
These two features work together to provide what Terracotta’s chief marketing officer Lynn Vojvodich calls a “go big, go fast,” approach, where access to more in-memory works with automated management controls to deliver better application performance, faster end user responses and even quicker application-to-application data integration for sharing on-premise or in the cloud.
“Traditional application architectures are stressed by exploding data growth. Our new offerings will give applications better access to more in-memory data, as well as provide IT automated tools to manage it.”
Mike Allen
Senior Director
Head of Product Marketing
Terracotta
Terracotta BigMemory Blasts Through in-Memory Limits
IT can’t always throw hardware with cheap memory at the problem because most software has limits as to how much in-memory resource it can access, Allen told IDN.
“Yes, it’s true. Prices for RAM are dropping 30% every 18 months, but the memory that’s addressable by Java remains limited,” Allen said. “In fact, the memory available to a JVM has hardly grown at all.” Even with improvements such as 64-bit technology and garbage collection most enterprises will not run a JVM with more than 4GB of heap, he added.
Terracotta’s latest upgrade to BigMemory crashes through this limit, allowing IT to access the 1 TB-sized RAM that now comes with the latest servers from IBM, Dell, HP and Oracle. “And this 1 TB access is per JVM, so just one Java application could access many times that, tens of TBs, at very quick micro-second speeds,” Allen added.
Under the covers, BigMemory is 100% Java and provides an in-process, off-heap cache that stores up to 1 TB of data closer to an application. These improvements of memory utilization work with both standalone and distributed caching.
BigMemory is extends the capabilities of open-source Ehcache, so it works on all popular JVM flavors (HotSpot, IBM JDK, and JRockit), app server and framework, Allen added. In turn, this drives flexibility so that devs can put their data where they want, and can even use it with new or existing 3-tier applications to offload data into memory without rewriting their code.
For implementation, BigMemory plugs into Terracotta’s commercial Enterprise Ehcache product with a few simple configuration changes. This is because despite the increase in size, there is no difference in how the cache is called, Allen added.
Terracotta’s Adds Automatic Resource Control to BigMemory
Terracotta engineers say that even with more in-memory, devs and operational IT also needs more control and management tools. Enter Terracotta’s new ARC.
“In ARC, you just tell the software how much memory you have and it handles it,” Allen said. “Other solutions require tuning, but we’ve put the intelligence directly into the software.”
To illustrate the value of ARC, Allen shares a colorful example:
“Without this control, it would be like flying a jumbo jet without an auto pilot,” Allen said. “Developers would need to do in-depth analysis to tune and re-tune and adjust to a variety of variables, such as customer usage. ARC does all that automatically, so like an autopilot you can just flip some switches and go fly fast and safe. “
For a more down-to-earth example, Allen turns to the familiar Java Pet Store. “If I want to allow 1000 pets in my cache I can do that,” he said. “But when I hit 1001 pets, one will get evicted. That means I have to iteratively go in and manage that. ARC avoids all that hand-tuning.”
Under the covers, ARC provides control for the in-memory layer via memory-based cache sizing at both the individual cache and grouped CacheManager levels. ARC also supports dynamic allocation of memory across multiple caches through automatic load balancing. The result is maximum efficiency with minimal tuning, Vojvodich said.
ARC can even help with performance management at very small cache sizes, Allen added. “People can get a lot of benefit from nonheap cache of 2 GB and laptops have 8GB of RAM,” he added, noting that every Microsoft PowerPoint can take control of a laptop’s entire memory with resource controls.
ARC is available in Terracotta’s commercial Enterprise Ehcache and BigMemory products, as well as the open source version of Ehcache.
Software AG Says In-Memory Has Broad Enterprise Architect Appeal
Terracotta’s innovative in-memory and control technologies are also capturing the attention of Software AG, its new parent, as a means to improve the lot of enterprise and integration architects, Allen added.
“Software AG has a lot of customers with high-value integration use cases that need more raid access to data,” he said. Among them: enterprise integration, cloud bursting, PaaS and SaaS integration. “These [latest] Terracotta technologies would be useful in all these types of profiles,” Allen said.
