2018-10-25-batch-split.md

Summary

Support BatchSplit feature that split one Region into multiple Regions at a time if the size is large enough. This includes modifications of both TiKV and PD. For TiKV, every round of split-check produces not only one split key but multiple split keys and change inner split related interface into batch style. For PD, add RPC AskBatchSplit and ReportBatchSplit to permit TiKV asking for region_id and peer_id in batch.

Motivation

Current split only split one Region at a time. It may be very slow when sequential write is too fast, namely, split speed can not keep up with write speed. Slow split can lead to large region. In this case, if a snapshot is triggered, it will occupy a lot of IO and make everything slow. Also, large region is hard for scheduling hotspot, so it makes performance even worse.

Detailed design

RPC interface

PD

service PD {
	// ...
	
    rpc AskSplit(AskSplitRequest) returns (AskSplitResponse) {
        // Use AskBatchSplit instead.
        option deprecated = true;
    }    
    rpc ReportSplit(ReportSplitRequest) returns (ReportSplitResponse) {
        // Use ResportBatchSplit instead.
        option deprecated = true;
    }
    rpc AskBatchSplit(AskBatchSplitRequest) returns (AskBatchSplitResponse) {}
    rpc ReportBatchSplit(ReportBatchSplitRequest) returns (ReportBatchSplitResponse) {}
}

message AskBatchSplitRequest {
    RequestHeader header = 1;
    metapb.Region region = 2;
    uint32 split_count = 3;
}

message SplitID {
    uint64 new_region_id = 1;
	repeated uint64 new_peer_ids = 2;
}

message AskBatchSplitResponse {
    ResponseHeader header = 1;
    repeated SplitID ids = 2;
}

message ReportBatchSplitRequest {
    RequestHeader header = 1;
	repeated metapb.Region regions = 2;
}

message ReportBatchSplitResponse {
	ResponseHeader header = 1;
}

Add AskBatchSplit to replace AskSplit , it is called when TiKV produces some split keys for one Region and asks PD to allocate new region_id and peer_id for that Region. split_count in AskBatchSplitRequest indicates the number of Region to be generated, and AskBatchSplitResponse returns all new allocated ids to TiKV.

Add ReportBatchSplit to replace ReportBatchSplit, it is called when TiKV finish splitting Region. ReportBatchSplitRequest takes all metas of new generated Region for PD to update PD's related information.

For compatibility issue, the old interface is not deleted but set to deprecated.

TiKV

message SplitRequest {
    // ...
    // Will be ignored in batch split, use `BatchSplitRequest::right_derive` instead.
    bool right_derive = 4 [deprecated=true];
}

message BatchSplitRequest {
    repeated SplitRequest requests = 1;
    // If true, the last region derive the origin region_id, 
    // other regions use new ids.
    bool right_derive = 2;
}

message BatchSplitResponse {
    repeated metapb.Region regions = 1;
}

enum AdminCmdType {
    // ...
    Split = 2 [deprecated=true];
    // ...
    BatchSplit = 10;
}

message AdminRequest {
    // ...
    SplitRequest split = 3 [deprecated=true];
    // ...
    BatchSplitRequest splits = 10;
}

message AdminResponse {
    // ...
    SplitResponse split = 3 [deprecated=true];
    // ...
    BatchSplitResponse splits = 10;
}

Add a new admin command type BatchSplit with related request and response. BatchSplitRequest wraps multiple SplitRequest along with right_derive which invalidates the right_derive in each SplitRequest.

When in rolling upgrade process, new TiKVs are mixed up with old TiKVs, so old command type Split still needs to be preserved.

Implementation in TiKV

How to produce multiple split keys

This part mainly focus on SplitChecker.

First of all, adjust trait to make it can return multiple split keys.

pub trait SplitChecker {
    // ...
    
    // before: fn split_key(&mut self) -> Option<Vec<u8>>
    fn split_keys(&mut self) -> Vec<Vec<u8>>;
    
    // before: fn approximate_split_key(&self, _: &Region, _: &DB) -> Result<Option<Vec<u8>>> 
    fn approximate_split_keys(&self, _: &Region, _: &DB) -> Result<Vec<Vec<u8>>> {
}

Then add one config batch_split_limit to limit the number of produced split keys in a batch. If it is unlimited, for once split check, it scans all over the Region's range, and in some extreme case this would cause performance issue.

Now we have four split-checkers: half, keys, size and table. SizeChecker and KeysChecker can be rewritten to produce multiple keys, and other checkers' logic stay unchanged.

The general logic of SizeChecker and KeysChecker are similiar, the only difference between them is one splits Region based on size and the other splits Region based on the number of keys. So here we mainly describe the logic of SizeChecker:

• before: it scans key-value pairs in a Region's range sequentially to accumlate their size as total_size and stops once the size reachs to region_max_size or scans to the end of range. If total_size is smaller than region_max_size at the end, checker wouldn't produce any split key; if not, it regards the very key at which total_size reachs to region_split_size as split key.
• after: it scans key-value pairs in a Region's range sequentially to accumlate their size as total_size and stops once the size reachs to region_split_size * (batch_split_limit-1) + region_max_size or scans to the end of range. During the scan process, it reocrds the key as split key every region_split_size, but after finishing scanning, it may discards the last split key if the size of rest Region doesn't over region_max_size - region_split_size. With this algorithm, if batch_split_limit is set to 1, TiKV can perfectly behave as before that split without batch.

Compatibility concern

The general process in raftstore changes a little, it mainly replaces Split with BatchSplit. But one thing should be noted, when rolling upgrade, PD version control will refuse AskBatchSplit request, thus split can't be performed during this process until all TiKV bump to new version. To let TiKV know whether AskBatchSplit fail for compatibility or not, we introduce a new error type for ResponseHeader :

enum ErrorType {
    // ...
    INCOMPATIBLE_VERSION = 5;
}

So once TiKV gets AskBatchSplitResponse with ErrorType::INCOMPATIBLE_VERSION, it uses original AskSplit instead of AskBatchSplit, and all following processes will degrade to original way. So original code path is not deleted.

Approximate split key

What we said above can ease the problem, however scanning a large Region can also consume a lot of time and CPU. Test shows that large Region can still easily show up even with batch split implemented, although split is speeded up.

When a Region becomes large enough, it's more practical to divide it into smaller chunks quickly. This can be achieved via size estimation, which can be calculated from SST properties. Although it may not be accurate enough, it's okay for a large Region.

So if the size of Region is larger than region_max_size * batch_split_limit * 2, TiKV uses approximate way to produce split keys. The approximate way is quite similar to the algorithm we describe above, but to estimate TiKV uses approximate size of the Region and the number of keys in the Region's range to calculate the average distance between two SST property keys, and produces a split key every region_split_size / distance keys.

Drawbacks

• When use approximate way, Region may split into several disproportion Regions due to size estimation.

Alternatives

None

Unresolved questions

A large Region is usually more emergent to be split, so we can change the split check queue from a naive FIFO queue to a priority queue so that large Region can be split early and quickly.