I assume you mean that when there are large number of concurrent requests, distributing ignite in a cluster distributes the load on multiple servers and hence reduces the average latency.
However, I am not load testing - I am comparing the latency in a single request to persist the blob data om two systems: a legacy system persisting the blob in an Oracle RDBMS and a new system (identical in RAM and cores) that uses IGFS to persist. thanx, /Kobe dsetrakyan wrote > How many Ignite nodes are you running in your cluster? Ignite is a > distributed system, so the more nodes you add, the faster it should get. > > On Wed, Feb 24, 2016 at 2:39 PM, Kobe < > rk_@ > > wrote: > >> Hello.. >> >> I am comparing the persistence of large blobs (megabytes) in Oracle >> relational database vs. IGFS (1.5.0.final, DUAL_ASYNC, backed by >> secondary >> filesystem) on a 64 bit, 16GB RAM, 8 core RHEL6 VM. >> >> I notice that the other parameters remaining constant, the time to >> persist >> the same blob in Oracle is a good 200 to 300ms faster than the time >> needed >> to persist the same blob on IGFS. >> >> I am trying to investigate this further. However, are there any obvious >> factors that may explain this paradox with IGFS? Is it possible that the >> buffering of data by Oracle server in its shared memory is more efficient >> that my attempt to persist the blob in IGFS? >> >> Has anyone else encountered this surprising result? >> >> thanks, >> >> /Kobe >> >> >> >> -- >> View this message in context: >> http://apache-ignite-users.70518.x6.nabble.com/Blob-persistence-performance-IGFS-vs-Oracle-tp3174.html >> Sent from the Apache Ignite Users mailing list archive at Nabble.com. >> -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Blob-persistence-performance-IGFS-vs-Oracle-tp3174p3177.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.
