rockeet

MGetCommand should use rocksdb MultiGet

1

now tendis implement MGetCommand one by one using rocksdb Get.

A proposal for Improving Key Format

2

Current Key Format is `SlotID | Type | DBID | PK | 0 | Version | SK | PK_LEN | Reserved`, the `PK_LEN` in suffix will harm some compression algo...

Feature Request: Iterator::Refresh() with a snapshot

4

Current Iterator::Refresh() does not support snapshot, we have no way to refresh an iterator to a specified snapshot, instead we create a new iterator, but creating new iterator is heavy....

MergeContext using unnecessary indirect pointers, and unnecessary std::string objects. This PR is just an optimization, no semantic changes.

IngestExternalFile put file on bottom level but on db_paths[0]

4

### Expected behavior `IngestExternalFile` should put file on db_paths according to level. ### Actual behavior `IngestExternalFile` put file on bottom level but on `db_paths[0]`. ### Steps to reproduce the behavior...

Report Fee Support

1

As a cloud native DB, distributed compaction needs fee charge, this involves some code changes.

Add Trace support to web REST api

1

```bash # StartTrace curl -d '{"file": "trace.txt", "filter": "kTraceFilterNone"}' "http://somehost:port/db/mydb?cmd=StartTrace" # EndTrace curl -d '{}' "http://somehost:port/db/mydb?cmd=EndTrace" ``` Start | End -----|----- StartTrace | EndTrace StartIOTrace | EndIOTrace StartBlockCacheTrace | EndBlockCacheTrace

Could use intel avx instructions when possible?

2

avx instruction has 256 bits instructions such as 256 bits popcount, this could improve the bit operations speed, could you add support for it?

Why do not use pg access methods for rocksdb

4

As I know, access methods is nearer to pg core than FDW, which implies there are less overheads. Is there any concerns?

This PR add two features to sysbench: 1. Read data from a file instead of generate random data by cmd option filename * This is to benchmark database by real...