Both Pipes and ASP.NET Core gRPC support local and remote IPC/RPC (with some platform limitations for gRPC)
When would I use one technology (Pipes) or the other (gRPC)?
Observations, thoughts and considerations I'm keeping in mind:
gRPC seems to be geared towards replacing WCF in some future iteration.
local deployments and with machine restrictions (running as non-admin/user, machine firewalls, different platforms/OS)
network traversal, and compatibility with same-machine -> multi-machine (frontend/backend arrays) for load and expansion
Spanning secure zones (where a Proxy is used, or other TLS cipher/order/registry setting) affects the ability for HTTP/2 to work
Pipes (named pipes?) have a different surface area and port (do they also use port 135, or NetBIOS over TCP (not sure of name))... how is it scanned and secured?
"memory mapped files" seem to be a challenge to get working, however it seems to work in ASP.NET Core with gRPC in the UDS configuration. Is this a correct inference?
Right now my scenario is to have two console apps communicate with each other, same machine or remote. Adding Asp.NET Core Web is an optional front end alternative for my scenario.
Simple IPC
Depends on how much communication is going to happen. If your communication is limited to simple collaborative signal passing or sharing some data between two processes you can safely use NamedPipeClientStream and NamedPipeServerStream on local system or local network but if you plan for the same on different systems then I would suggest using TcpClient and TcpListener.
Comprehensive IPC
WCF or now its replacement gRPC is for scenario where a complete API/Framework need to be executed remotely. For example I have an entire library of classes which I need to call from a different process (which mostly run on a different system); in that case gRPC kind of solutions make more sense.
Only you can decide.
This is a design decision which is highly unique for your application; your future plans and your system environment and any third person can only give you clues but ultimately you are the only person who can make the right decision.
Related
Just a technology update, now that .NET 4.0 is out.
I write an application that communicates to the server through what is basically a message bus (instead of method calls). This is based on the internal architecture of the application (which is multi threaded, passing the messages around).
There are a limited number of messages to go from the client to the server, quite a lot more from the server to the client. Most of those can be handled via a separate specialized mechanism, but at the end we talk of possibly 10-100 small messages per second going from the server to the client.
The client is supposed to operate under "internet conditions". THis means possibly home end users behind standard NAT devices (i.e. typical DSL routers) - a firewalled secure and thus "open" network can not be assumed.
I want to have as little latency and as little overhad for the communication as possible.
What is the technologally best way to handle the message bus callback? I Have no problem regularly calling to the server for message delivery if something needs to be sent...
...but what are my options to handle the messagtes from the server to the client?
WsDualHttp does work how? Especially under a NAT scenario?
Just as a note: polling is most likely out - the main problem here is that I would have a significant overhead OR a significant delay, both aren ot really wanted. Technically I would love some sort of streaming appraoch, where the server can write messags to a stream while he generates them and they get sent to the client as they come. Not esure this is doable with WCF, though (if not, I may acutally decide to handle the whole message part outside of WCF and just do control / login / setup / destruction via WCF).
For bidirectional communications, your best bet is NetTcpBinding, rather than the http bindings, if they're available.
This has the advantage of only requiring that the client can initiate a connection with the server.
I would go with Windows Azure Service Bus. See my answer in the following question:
WCF, 4.0, Bidirectional
Take a look at Windows AppFabric, good place to start is Here. It fundamentally wraps up WCF and WF into an application server, with WCF activation supported through WAS. Its where I would host this type of app. It offerd full duplex connection orientated, p2p or sessions between client and server. Don't confuse the Windows appfabric with Azure appfabric, (formely called Azure Service Bus).
As regards bindings above, both NetTcpBinding and WsDualHttp offer callbacks, but the ws binding you get a lot for your cash, especially if it's a mixed programming environment and you have to flatten the wsdl to make interop work. I also think that WsDual is easier on routers traversal, although I understand talking to friends, that Windows AppFabric mitigates this, with new Relay Services, (which i've not seen, and I think have now been renamed).
Hope that helps.
We are working on developing a Java EE based application. Our application is Java 1.5 compatible and will be deployed to WAS ND 6.1.0.21 with EBJ 3.0 and Web Services feature packs. The configuration is currently one cell with two clusters. Each cluster will have two nodes.
Our application, or our system, as I should rather say, comes in two or three parts.
Part 1: An ear deployed to one cluster that contains 3rd party vendor code combined with customization code. Their code is EJB 2.0 compliant and has a lot of Remote Home interfaces.
Part 2: An ear deployed to the same cluster as the first ear. This ear contains EBJ 3's that make calls into the EJB 2's supplied by the vendor and the custom code. These EJB 3's are used by the JSF UI also packaged with the EAR, and some of them are also exposed as web services (JAX-WS 2.0 with SOAP 1.2 compliance) for other clients.
Part 3: There may be other services that do not depend on our vendor/custom code app. These services will be EJB 3.0's and web services that are deployed to the other cluster.
Per a recommendation from some IBM staff on site here, communication between nodes in a cluster can be EJB RMI. But if we are going across clusters and/or other cells, then the communication should be web services.
That said, some of us are wondering about performance and optimizing communication for speed of our applications that will use our web services and EJB's. Right now most EJB's are exposed as remote. (and our vendor set theirs up that way, rather than also exposing local home interfaces). We are wondering if WAS does any optimizations between apps in the same node/cluster node space. If two apps are installed in the same area and they call each other via remote home interface, is WAS smart enough to make it a local home interface call?
Are their other optimization techniques? Should we consider them? Should we not? What are the costs/benefits? Here is the question from one of our team members as sent in their email:
The question is: Supposing we develop our EJBs as remote EJBs, where our UI controller code is talking to our EXT java services via EJB3...what are our options for performance optimization when both the EJB server and client are running in the same container?
As one point of reference, google has given me some oooooold websphere performance tuning documentation from 2000 that explains a tuning configuration you can set to enable Call By Reference for EJB communication when they're in the same application server JVM. It states the following:
Because EJBs are inherently location independent, they use a remote programming
model. Method parameters and return values are serialized over RMI-IIOP and returned
by value. This is the intrinsic RMI "Call By Value" model.
WebSphere provides the "No Local Copies" performance optimization for running EJBs
and clients (typically servlets) in the same application server JVM. The "No Local
Copies" option uses "Call By Reference" and does not create local proxies for called
objects when both the client and the remote object are in the same process. Depending
on your workload, this can result in a significant overhead savings.
Configure "No Local Copies" by adding the following two command line parameters to
the application server JVM:
* -Djavax.rmi.CORBA.UtilClass=com.ibm.CORBA.iiop.Util
* -Dcom.ibm.CORBA.iiop.noLocalCopies=true
CAUTION: The "No Local Copies" configuration option improves performance by
changing "Call By Value" to "Call By Reference" for clients and EJBs in the same JVM.
One side effect of this is that the Java object derived (non-primitive) method parameters
can actually be changed by the called enterprise bean. Consider Figure 16a:
Also, we will also be using Process Server 6.2 and WESB 6.2 as well in the future. Any ideas? recommendations?
Thanks
The only automatic optimization that can really be done for remote EJBs is if they are colocated (accessed from within the same JVM). In that case, the ORB will short-circuit some of the work that would otherwise be required if the request needed to go across the wire. There will still be some necessary ORB overhead including object serialization (unless you turn on noLocalCopies, with all the caveats it brings).
Alternatively, if you know that the UI controller is colocated, your method calls do not rely on parameter or return value copying, and your interface does not rely on the exception differences between local and remote views, then you could create and expose a local subinterface that will be much faster than remote access through the ORB.
What would be reasons that you want to host a wcf service in a windows service and not in IIS?
One reason is that IIS6 only supports bindings based on HTTP. If you want to use TCP, MSMQ, etc., then you need to host in a separate program.
When hosting in IIS you are only allowed to bind to a single port per a base address, in each web site (Meaning you cant specify two bindings with different ports since you can only use a single port, or endpoints that use different ports)
You can only use a single base address in IIS, the only way around this is deploying multiple versions of the same project in different websites (yuck)
The IIS process must recycle eventually, and when it does it dumps everything and restarts, which is good alot of the time since memory is freed and trapped resources are released, but when using singletons this can have an undersired effect depending on your code
[edit] : more points
In a standard setup your worker process always have 2GB Virtual memory available (no matter if you have 1, 2 or 4GB physical memory in the machine).
Freedom. You as the developer don't need someone to administer the box
Sometimes IIS6 is really just overkill
You are using it as interprocess communication conduit
You wish to declare all of the bindings in code. This is far less confusing and more powerful than the xml config files that seem to be all the rage. I can't envision many scenarios where I would want a non-programmer messing with bindings. The xml approach is fine for prototyping and systems that need to be highly dynamic, but overall I don't think its a good idea
We're planning a system running on Windows/.Net 3.5 that has a number of "services" that need to run in the background. Some will be active all of the time, but some will only be called occassionally and can be stood-up on demand.
As far as I can see, my options are:
Windows Services - always running(?)
IIS hosted something - called on demand
COM+/ .Net Enterprise Sevices - most complex option, but most powerful?
Distributed transactions is not a requirement, these are mainly computation engines, rather than transaction processors.
Does anyone have any experience of working with all of these and what further pros & cons can be claimed for each technology?
EDIT
Is suppose there are multiple ways of hosting code in IIS, web services, WCF (as pointed out below), any others? Relative pros/cons?
WCF feels like the right way to go. There are still many choices to make. WCF provides a number of communication mechanisms and hosting environments:
WCF combines the following technologies under one set of APIs-
ASMX;
WSE;
Remoting;
COM+;
MSMQ.
So for instance you can use persistent messages from MSMQ for occassionaly connected clients or standard XML encoding SOAP messages over an HTTP transport layer. You can also use new features in 3.5 like binary encoding of XML or JSON encoding over HTTP.
Hosting environments include:
Console applications
Windows services
WCF services inside IIS 7.0
and on Windows Vista or Windows Server 2008 you can use WAS (Windows Activation Services) to host WCF services.
Different hosting environments have pros and cons. I suggest you look at MSDN for more details (e.g. http://msdn.microsoft.com/en-us/library/bb332338.aspx).
Because WCF encompasses a lot of functionality it is more difficult to learn than any one of the technologies it replaces. I still think it pays for itself in the long run.
It depends on what the software will do, and how (and if) users or systems need to interact with it. Depending on those things, there may be one more, often overlooked, option: set it up as a scheduled task. This is often a very good alternative to a windows service, if the software is of the kind that will act on certain time intervals (check for a change in a database, act on the changed data and send it somewhere, for instance).
If you will have other systems talking directly to your software, I would imagine that a WCF application hosted in IIS would be a rather straighforward way. We use both those approaches in my current assignment; WCF services for looking up and storing data, and scheduled tasks for data calculations that run on a regular basis.
The scheduled task has one upside compared to the others in one specific field; it uses system resources only when running.
You mentioned starting up a process "on demand". WAS - Windows Activation Service, or sometimes called Windows Process Activation Servvice, though it is never abbreviated "WPAS" - is the thing inside Windows that provides on-demand process activation. The way it works - when a message arrives, WAS can start a worker process to handle the message. WAS was, prior to IIS7, fairly tightly integrated into IIS. It was used primarily to activate processes that did web work - like an ASP.NET worker process. With IIS7, WAS is generalized so that it can activate worker processes based on non-HTTP as well as HTTP messages. If you write your app to receive messages through WCF, you can get activation essentially "for free". That applies if it is HTTP, TCP, MSMQ; SOAP or otherwise.
The key thing with this on-demand startup though, is that it is tied to the communication. In fact the process lifecycle model for WAS is tied to communication as well. By default if there are no incoming messages after a while, the process will be shut down by WAS. That may or may not be what you want.
As for process hosting - COM+ offers a hosting environment but it is primarily intended for use as a host for processes that communicate. This may not be the perfect fit for you.
If you have compute engines, you may just want to run a Windows Service. A service like that can be started and stopped either administratively or programmatically. In the latter case, you could imagine a WAS-activated worker process programmatically starting a windows service.
You could also imagine writing a simple Windows Service that watches a location (filesystem, message queue, etc) for a message, and when that file or message arrives, the Windows Service starts up a compute engine process, which itself is NOT a Windows Service, but is just a process.
Speaking of MSMQ - That is basically the same model as MSMQ triggers. You can configure MSMQ to start a process when a message arrives on a particular queue.
There are lots of options.
I am looking at using MSMQ as a solution to do asynchronous execution in my upcoming project. I want to know the differences between using WCF and frameworks like MassTransit or even hand written MSMQ client to place/read task off MSMQ.
Basically the application will be several websites (internal through LAN or external through the Internet) reading/writing data through a service layer (be it WCF or normal web service). Then this service layer will do one of two things: 1. write data to database 2. and/or trigger the background process by placing a message in the queue. 3. obviously it can also retrieve data from database. The little agent (a windows service) on the other side of the queue will monitor the queue and execute based on the task command.
This architecture will be quite easy to scale (add more queues and agents) and easy to implement compared to RPC or distributed execution or whatever. And the agent processing doesn’t need to be real time. And the agent and service layer are separate applications except they share the common domain objects and Repositories etc.
What do you think? Architecture suggestions for the above requirements are welcomed. Thank you!
WCF adds an abstraction over MSMQ. In fact, once you define compatible contracts (operations must be OneWay), you can switch out MSMQ in the config, transparently. (For instance, you could switch to normal HttpWS or a NetTcp binding.)
You should evaluate the other WCF benefits, like security and so on, to see how those fit in with your needs. Again, they should be reasonably transparent of the fact you're using MSMQ underneath. For instance, adding SOAP security and so on should "just work", independent of using MSMQ.
(Although, IIRC, you still need to login to the desktop on each machine that uses MSMQ, with the service account that will use MSMQ, to generate the certificate in the machines local profile. And then, it doesn't work very well from IIS6, since user profiles aren't loaded. A real pain in general, but nothing to do with WCF specifically.)
Apart from that:
Have you looked at SQL Server Service Broker? After using MSMQ + WCF and SSSB, I think that SSSB is vastly easier to configure and manage. SSSB works with T-SQL commands over any SQL client (I use it from Mono, on Linux, with transactions). It'll also give you transactional send/receive, even remotely (I think MSMQ 4 now allows this). It really takes a lot of the pain away from message queuing, and if you're using SQL Server already...
SSSB is often overlooked since the SQL Management Studio doesn't have GUI designers for it all, but it isn't hard and is a great option. The one downside is that if you want local send capability (i.e., queue message when network is down), you'll need to run a local SQL Express instance.
Your architecture seems sound and reasonable. However you should consider using the WCF net MSMQ transport over hand coded MSMQ classes. WCF wraps this common functionality into a nice programming model. Also I believe there is some improvements in the protocol used by wcf compared to basic System.Messaging
Have a look at the value-add over plain MSMQ:
http://readthedocs.org/docs/masstransit/en/latest/overview/valueadd.html
In summary, you get a lot of messaging concepts clearly presented in the API with MassTransit; to an extent you wouldn't have if you hand-coded it or used WCF.