Why learn service grid istio

Istio Is running on the kubernetes A service grid on (Service Mesh, Third generation microservice Architecture ), Greatly expanded kubernetes The function of , It can be easily achieved than kubernetes Flow management with finer granularity .

When we are directly in kubernetes When deploying an application , Sometimes it is difficult to realize some special needs , Like the Canary release （ Also known as grayscale publishing ）.

Canary publishing works like this , We already have an old version of the application here called v1, Then we will release a new version of the application, which is called v2. But now I don't want all clients to access v2, I just want to open some clients for testing in a small range .

So we can set 90% Traffic forwarded to v1,10% Traffic forwarded to v2. If this 10% User feedback v2 Not bad , Then we can expand our visit v2 Of traffic , Narrow access v1 Of traffic .

And so on , Until all the traffic is gradually transferred to v2 above .

without istio The way of implementation is like this , Create two deployment, One is v1 One v2. They created pod Have the same label for service relation . First v1 Set up 3 Copies ,v2 Set up a copy , such 75% All traffic is forwarded to v1 Yes ,25% Traffic forwarded to v2 On .

Then we can shrink v1 Number of copies , increase v2 To modify the flow direction , Here's the picture .

But the flow control is too rough , And it is too troublesome to maintain .

But we are istio It is very convenient to realize . The following is taken from a VirtualService（ Referred to as vs） Part of the code in , We changed it weight You can modify the flow direction .

    - destination:
        host: svc1
        subset: v1
      weight: 65
    - destination:
        host: svc1
        subset: v2
      weight: 35

Here is designated to pod1 Of traffic 65%, Go to pod2 Of traffic 35%, Then look at the flow direction through the figure below .

And then we modify it again weight.

    - destination:
        host: svc1
        subset: v1
      weight: 35
    - destination:
        host: svc1
        subset: v2
      weight: 65

Set here to pod1 go 35% Of traffic , Go to pod2 go 65% Of traffic , And then look at the picture below .

We just need to change the weight You can easily forward to pod Of traffic . In addition to the Canary release above , We want to achieve blue-green deployment 、A/B test 、 Conversation retention 、 Traffic mirroring and other functions can be easily realized , therefore istio Its strength lies in its traffic management .

If we want different clients （Android、windows、firefox、chrome etc. ） Visit different pod On ,istio It can also be easily realized . in addition istio It also provides strong observability , Because we installed kiali, You can also easily see the flow direction .

istio It also provides us with more powerful security management , Including passage AuthorizationPolicy Do more detailed authorization management 、pod Between mTLS Certification, etc , Greatly improved pod The security of .

With the advancement of open source and cloud computing , As the core technology, cloud native microservices have maintained rapid growth ; With the maturity of microservice Technology , Began to penetrate all walks of life . On the one hand, labor costs are rising , It is imperative to adopt micro services to improve R & D efficiency .