On 10/31/18 8:16 AM, Sven Van Caekenberghe wrote:
Hi Jimmie,
On 31 Oct 2018, at 14:01, Jimmie Houchin <[email protected]> wrote:
Hello,
I am writing an app which accesses a REST api. The documentation says this.
Connection Limits
To provide equal resources to all clients, we recommend limiting both the
number of new connections per second, and the number of requests per second
made on a persistent connection (see above/below).
For new connections, we recommend you limit this to twice per second (2/s).
Establishing a new TCP and SSL connection is expensive for both client and
server. To allow a better experience, using a persistent connection will allow
more requests to be performed on an established connection.
For an established connection, we recommend limiting this to one hundred per
second (100/s). Please see following section on persistent connections to
learn how to maintain a connection once it is established.
HTTP Persistent Connection
This keeps the underlying TCP connection active for multiple
requests/responses. Subsequent requests will result in reduced latency as the
TCP handshaking process is no longer required.
If you are using an HTTP 1.0 client, ensure it supports the keep-alive
directive and submit the header Connection: Keep-Alive with your request.
Keep-Alive is part of the protocol in HTTP 1.1 and enabled by default on
compliant clients. However, you will have to ensure your implementation
does not set the connection header to other values.
I do not know how to reuse a connection.
I see in ZnClient #canReuseConnection and #connectionReuseTimeout.
Is the timeout from connection time or last request or access time?
Does that handle everything automatically for me?
I will be accessing this api multiple times a minute, sometimes multiple times
a second for almost all the time.
How long can I keep a persistent connection open?
What are best practices?
Any help and wisdom greatly appreciated.
Thanks.
Jimmie
Yes, Zn fully and automagically supports persistent connections.
In a first approach, just create one ZnClient instance, configure it with your
host, etc, and then keep on reusing this instance for each (subsequent)
request. Zn will keep client side connections open indefinitely. But a (busy)
server will kill idle connections for sure. ZnClient will always try to connect
again, before giving an error, so it effectively reconnects.
You can look at client side logging by inspecting
ZnLogEvent announcer
to figure out what is happening.
One warning though: if your app is multi user / multi threaded, you will
probably need an http client instance per session. It is possibly necessary to
prevent mixing up authentication info as well.
HTH,
Sven
Thanks Sven. I thought that would be how it worked.
You write beautiful software. I love to try to learn from what you write.
Thanks for the pointer.
Jimmie
