Aspnetcore.docs: Drop 'reverse proxy' language

[guardrex]

https://github.com/aspnet/Docs/pull/9463#issuecomment-436420203

We'll be dropping 'reverse proxy' language where possible per the engineering guidance. Replacement text will go along the lines of ...

the ANCM allows an app to run behind IIS

[Rick-Anderson]

@Tratcher 'reverse proxy' is widely used. If we want to use a more precise term, we should keep reverse proxy. For example:

the ANCM allows an app to run behind IIS (commonly called a reverse proxy)

that way customers searching on ASP.NET Core reverse proxy can find what they're looking for.

cc @tdykstra @scottaddie @jkotalik

[guardrex]

widely used

... 88 references across 16 markdown files. IIRC, it was ~97 refs, but we removed a few on the doc-a-thon PR.

[Tratcher]

@jkotalik, @guardrex what's wrong with the reverse proxy language? That's what ANCM out-of-proc is. Any specific examples where you think it's misleading? Or is this more about cutting the verbosity?

@shirhatti

[guardrex]

@Tratcher - @jkotalik will need to say ... I'm just following orders.

WRT in-process, can you explain a bit more? Definitions of "reverse proxy" seem to pertain to the number and arrangement of servers for request processing, not how the servers' code is executed on the host (i.e., what process they're running in). Is Kestrel no longer a "server" when hosting in-proc?

[Tratcher]

When ANCM hosts ASP.NET Core in-proc it doesn't use Kestrel, it passes requests directly to the app. In this case it's no longer acting as a proxy, but the direct server (i.e. there's only one server here).

[guardrex]

Oh, yes, of course ... IISHttpServer ... thanks for reminding me.

However, does the question remain?

There's IIS and IIS HTTP Server† (IISHttpServer) in-proc ... that's not considered two servers?

[†"IIS HTTP Server" – I guess that's how we'll write it into content.]

[Tratcher]

It's no more two servers than HttpSysServer is two servers. They are certainly two layers, but not two independent servers.

[jkotalik]

Any specific examples where you think it's misleading?

@Tratcher there isn't anything wrong with the reverse proxy language, per say. The issue is more that if we want to push ANCM InProc as the way to hosting behind IIS from 2.2 and beyond, we should check any place that contains the language "Reverse Proxy" and fix modify it accordingly (whether it be with monikers or changing the language in general).

[Tratcher]

Ah, generalizing to account for in-proc where appropriate makes sense.

[guardrex]

I believe I'm in good shape now to perform the update.

I'll work this issue after the current batch of IIS-related PRs are merged.

[tdykstra]

@guardrex In the docs where we specifically talk about out of proc use would it be helpful to keep the reverse proxy language and explain the difference between that and in-proc? Or do we already do that?

[guardrex]

@tdykstra I think we did a good job for the 2.2 updates; so in those spots, I think we're ok. Those references are tied into the hosting model description and seem clear to me. For example ...

• https://docs.microsoft.com/aspnet/core/fundamentals/servers/aspnet-core-module?view=aspnetcore-2.2#aspnet-core-module-description
• https://docs.microsoft.com/aspnet/core/host-and-deploy/aspnet-core-module?view=aspnetcore-2.2#hosting-model

... but I'll look again just to make sure.

Idk on the other _~70+ references_. I'll check all of them.

[guardrex]

@shirhatti I'm going to work the "reverse proxy" language (this issue) now. I'm holding off on the Servers TOC node work for your signal to proceed. Let me know if your schedule changes that affects that work prior to 12/31 ... I'd like to get that one done while I'm still on-the-clock here.