Today, Solace and GigaSpaces announced a new joint offering for fast and efficient data synchronization between distributed spaces. We call it “space bridging” and it addresses key challenges for a variety of the most vexing large-scale app server or high performance computing (HPC) problems when real-time data is required in multiple geographic locations.
GigaSpaces is most commonly used when interactions with a data set are so frequent or so numerous, that it is too slow, or too burdensome on the database to be fetching/updating over and over using traditional data access. So GigaSpaces steps in and provides a clustered in-memory solution that allows the reads to happen in memory, and the updates to be cached for more efficient writing to the permanent data store. This is important, much needed functionality for many high performance applications.
While this works great with closely coupled LAN-based computing clusters, what happens when you have an application with users in New York, London and Tokyo all sharing the same information? How much lag can the application tolerate as data is changing in one of these locations, but is not yet available in the others? Is one location the master, and the others just replicants, or do they all need to be able to capture and resolve conflicts between updates? How many updates per second can realistically be synchronized between geographically distributed spaces?
These are the kinds of questions that space bridging helps address. GigaSpaces has a widely-used API for sharing information between spaces. Solace has the world’s fastest WAN-optimized, guaranteed delivery middleware. Solace and GigaSpaces got together and integrated the technologies to improve WAN-based space replication by 10 or more times with facilities to deal with mirroring and update conflict resolutions. In fact, where bandwidth is not a limiting factor, wide area update throughput can be MUCH higher than 10 times. Even where WAN bandwidth is tight, sending less information and/or using compression can squeeze more performance out of the same resources.
You’ll be reading much more about this solution here and on the GigaSpaces blog in the coming weeks. For today, we are pleased to announce that this new capability is here and welcome our friends at GigaSpaces and distributed systems architects that work with their system to the discussion.