Controller poor performance with large amount of hosts.

[zzjin]

If you are reporting any crash or any potential security issue, do not
open an issue in this repo. Please report the issue via ASRC(Alibaba Security Response Center) where the issue will be triaged appropriately.

• I have searched the issues of this repository and believe that this is not a duplicate.

Ⅰ. Issue Description

When one k8s cluster have thousands of ingress with different hostnames,higress-controller will use large amount of cpu and memory, and result in newly created or modified ingress takes about ten seconds to become accessible, during which time it may show 404, 500, or 503 errors.

Ⅱ. Describe what happened

The newly created or modified ingress can be accessible under seconds.

Ⅲ. Describe what you expected to happen

It is needed by tens seconds or more to reconcile ingress.

[johnlanni]

higress currently depends on istio at the bottom layer. When there are many domain names, istio generates LDS (containing a large number of filter chains), which consumes a lot of CPU.

By adding the cache of LDS, this problem can be solved.

istio currently does not have an xds cache for LDS, it can be extended on the higress fork.