refactor(clustering/rpc): cache the result of parse_method_name()

[chronolaw]

Summary

Using lru_cache to improve the performance of pares_method_name().

KAG-4441

Checklist

• The Pull Request has tests
• A changelog file has been created under changelog/unreleased/kong or skip-changelog label added on PR if changelog is unnecessary. README.md
• There is a user-facing docs PR against https://github.com/Kong/docs.konghq.com - PUT DOCS PR HERE

Issue reference

Fix #[issue number]

[ADD-SP]

The method should not be very long, I'm wondering if the lrucache is significantly faster than this simple string operation.

I'm trying to avoid meaningless memory consumption.

[chronolaw]

Yes, I also notice that, need more discussion.

[chronolaw]

I did a simple perf test:

-- lru cache
local function parse_method_name1(method) ... end

-- no cache
local function parse_method_name2(method) ... end

local COUNT = 100 * 1000

ngx.update_time()
local t = ngx.now()

for i = 1, COUNT do
  parse_method_name1("kong.rpc.v1.test")
end

ngx.update_time()
print("v1: ", ngx.now() - t)

local t = ngx.now()

for i = 1, COUNT do
  parse_method_name2("kong.rpc.v1.test")
end

ngx.update_time()
print("v2: ", ngx.now() - t)

The result is：

resty ./t.lua
v1: 0.002000093460083
v2: 0.0079998970031738

So cache is faster than calculation.