https://github.com/robaho/keydbr

remote protocol layer for keydb using gRPC. deprecated. use robaho/leveldbr
https://github.com/robaho/keydbr

database grpc key-value keydb leveldb lsm lsm-tree stream

Last synced: 3 months ago
JSON representation

remote protocol layer for keydb using gRPC. deprecated. use robaho/leveldbr

• Host: GitHub
• URL: https://github.com/robaho/keydbr
• Owner: robaho
• License: gpl-3.0
• Created: 2018-09-14T04:03:09.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2022-09-13T01:11:11.000Z (over 2 years ago)
• Last Synced: 2024-06-20T15:54:04.070Z (8 months ago)
• Topics: database, grpc, key-value, keydb, leveldb, lsm, lsm-tree, stream
• Language: Go
• Homepage:
• Size: 52.7 KB
• Stars: 8
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: readme.md
  • License: LICENSE

Awesome Lists containing this project

README

        
## keydbr is deprecated. Use [robaho/leveldbr](http://www.github.com/robaho/leveldbr) which is stable and uses an api based on Google's LevelDB.

**About**

Remote access to keydb database. Uses Google protobufs and gRPC.

**Notes**

There is a single gRPC stream per open database,

upon which all requests are multiplexed. The stream is shared but requests are completed synchronously (since the server processes an inbound message synchronously) but this may change in the future to allow overlapping requests - that is, asynchronous handling

by the server. For best performance, multiple connections should be made to the server, rather than sharing a database connection.

**To Use**

go run cmd/server

There is a sample command line client in cmd/client which uses the client API.

**Performance**

Using the same 'performance' test as keydb, but using the remote layer:


insert time  1000000 records =  3080 ms, usec per op  3.08019

close time  1700 ms

scan time  2129 ms, usec per op  2.129979

scan time 50%  1468 ms, usec per op  2.93625

random access time  83.98727 us per get

close time  1000 ms

scan time  1963 ms, usec per op  1.963984

scan time 50%  930 ms, usec per op  1.861844

random access time  82.95168 us per get



**TODOs**

Implement "read ahead" for more efficient lookup over the network