CMR-9498 Changed metadata cache from in-memory search only to redis where bootstrap loads the data in non dev environments

[eereiter]

and search will load the data for development environments using in memory database. I pushed a lot of code to common so that both bootstrap and search can use the same functions when possible.

[codecov-commenter]

Codecov Report

Merging #1993 (a220aec) into master (de4c9d6) will decrease coverage by 0.18%.
Report is 3 commits behind head on master.
The diff coverage is 20.00%.

@@            Coverage Diff             @@
##           master    #1993      +/-   ##
==========================================
- Coverage   55.26%   55.08%   -0.18%     
==========================================
  Files         992      995       +3     
  Lines       63036    63255     +219     
  Branches     1561     1566       +5     
==========================================
+ Hits        34836    34847      +11     
- Misses      26643    26846     +203     
- Partials     1557     1562       +5     

Files	Coverage Δ
...h/data/metadata_retrieval/metadata_transformer.clj	41.89% <ø> (ø)
...pp-lib/src/cmr/common_app/services/kms_fetcher.clj	53.19% <50.00%> (ø)
...utils-lib/src/cmr/redis_utils/redis_hash_cache.clj	53.57% <0.00%> (ø)
...results_handlers/generics_json_results_handler.clj	4.05% <0.00%> (ø)
...h/services/humanizers/humanizer_report_service.clj	30.76% <0.00%> (ø)
...search/services/parameters/provider_short_name.clj	36.95% <0.00%> (ø)
...app-lib/src/cmr/common_app/services/cache_info.clj	43.58% <0.00%> (-1.15%)	⬇️
...rc/cmr/common/cache/single_thread_lookup_cache.clj	89.28% <91.66%> (-0.72%)	⬇️
search-app/src/cmr/search/system.clj	18.29% <0.00%> (-0.23%)	⬇️
...pp/data/metadata_retrieval/revision_format_map.clj	57.60% <63.63%> (ø)
... and 9 more

[daniel-zamora]

Has changing this cache to use redis been evaluated for performance? In production, we only use a single node redis cluster, will this need to be expanded(how large will the cache be and how often does it get accessed)?

[eereiter]

Has changing this cache to use redis been evaluated for performance? In production, we only use a single node redis cluster, will this need to be expanded(how large will the cache be and how often does it get accessed)?

No, we have not done performance tests with changing this cache from memory to redis. In production we have increased the redis nodes from 1 primary to 3; 1 primary and 2 replicas. This cache is used to satisfy collection umm_json search requests - all other concept types still go to the database - that code hasn't changed. I have read that we should get micro-second performance out of redis, but I haven't tested it to see if that is true.