On Mon, 2015-05-04 at 10:03 +0100, Vijay Bhoomireddy wrote: > Just wondering whether there is a provision to store SolrCloud index data on > Amazon S3? Please let me know any pointers.
Not to my knowledge. >From what I can read, Amazon S3 is intended for bulk data and has really poor latency. For example: https://engineering.gnip.com/s3-read-performance/ puts the latency of a 1KB-request at 125ms. Interestingly enough it seems that multi-threading helps tremendously in this scenario. Standard Lucene/Solr does not take advantage of this. One could conceivably over-provision reads, but it seems like a lot of work for a special case. - Toke Eskildsen, State and University Library, Denmark
