I am far from an Ignite expert but I believe the separate Grids is Best and I use that myself.
Mike (MO) Oliver's iPhone 09479927462 > On 18 Dec 2017, at 6:34 AM, Raymond Wilson <[email protected]> wrote: > > I have a use case I’d like to explore with Ignite: > > We have a data ingest pipeline that transforms in bound data into a large > collections of mutable records (stored in an Ignite cache). Changes to this > mutable data is relatively frequent and imposes significant IO updating > changes. From those records an immutable projection is computed which is much > smaller from a space perspective (stored in a separate Ignite cache). > > I’d like to have a set of servers responsible for processing and storing the > mutable data set that is independent from a set of servers storing the > immutable data projection and servicing requests against it. In this way I > don’t compromise the servers handing requests with the IO and storage > overhead of the mutable data – they care only about the small immutable data. > > Currently you can configure persistence (or not) on a cache by cache basis, > but not storage (which is configured on a per grid basis in the > IgniteConfiguration). > > Assuming that is correct, what is the easiest way to achieve the data > separation as described above > > Two obvious choices are: > > Instantiate two grids and use the persistence configuration to establish the > separate sets of servers. This seems simple to set up and understand. > Use a single grid, but use affinity mapping to route different caches to > different sets of servers. This seems more complicated. > > Thanks, > Raymond. >
