Zowe: What is Zowe

At one level Zowe allows people running on their works stations to access z/OS without having to logon to TSO and use ISPF. This can be done using

  • A browser
  • A scriptable command level interface(s)
  • Plug in to VSCode, so you can do all your work using the IDE.
  • You own code using a REST API using URLs and data in JSON format.

What is unique about Zowe?

There are other products which provide similar function. For example z/OSMF (z/OS Management Facility) provides many of the same facilities. Zowe uses z/OSMF for a lot of the function.

I’ll give a bit of history to show how Zowe and z/OSMF fit in today’s environment.

The 1970’s

In the the 1970’s a client machine would connect to z/OS quite likely using a proprietary interface. There may be one or just a few (for availability) back end servers . A typical client might connect to the server in the morning, and stay connected all day until the client machine was shutdown. The cost of using the networks was high, and a typical transaction had several flows to and from the server, sending updates (rather than the whole transaction data – see below).

One of the problems with this model is if you start another server mid-morning, it may get very few connections because the client connected to the server first thing, and might only go to the new server if they had to restart.

Another problem is that the client only signed on once a day, and if a userid was revoked it would still be in use till the client shut down.

Both of these problems can be solved by having the clients periodically disconnect at the end of a transaction and reconnect.

Today…

The architecture has matured. The web browser is used as a front end for much of the transactions. A typical request is now a url like

https://bigshop.co.uk:5555/sales?part=123456,name=ColinPaice


Where

  • https: is the protocol, another protocol could be ftp
  • bigshop.co.uk: is the IP address (or the name which maps to an IP address)
  • 5555: the port on the server
  • sales: this is the transaction
  • ?: splits the transaction from the data
  • part=123456,name=ColinPaice: this is the data passed to the transaction.

When this request flows to the server there may be a software router in z/OS which says “if this is a new session request – then send it to the lightest used server”. This gives load balancing. If you issue the same request multiple times it may go to different servers each time.

The request gets to a server machine. Several instances of an application can be running listening on the port 5555. Again this provides workload balancing.

One shot request

A request can be one-shot – start a session, authenticate, do something, get a response back – end. This provides a highly available scalable solution. You can take servers in and out of commission and work will execute.

Conversation request

A session can be long lived, where there are many flows within a session. For example list all data set, display this member etc. This does: start a session, authenticate, have a conversation end.

When the server responds to the request it sends back the IP address for future traffic in the session, and the session specific port. When the client sends data within the session, it goes to this partner session.

Authentication

Authentication can be expensive. For example using TLS to provide secure network flows, requires several network flows. Using a certificate or userid and password can be expensive. TLS has a Session resumption or fast reconnect. If you disconnect and reconnect again the client can send a token, the server can validate it, and if it is valid bypass some of the set up.

To reduce the costs at the application to application level, you can use an authentication token. Once you have authenticated, you are given an encrypted token. This token contains your userid and other information, and is valid for a time period ranging from minutes to hours. If the token expires you have to re-authenticate to get a new token. This token may be valid across server instances on one machine, and may be valid on servers on different systems.

How do you issue a request?

You can write your own program to establish a TCP/IP session to the server and send and receive data. There are several tools to help you, including cURL, openssl client, Python, Java and the Zowe client.

Many services are REST services, where the server has no saved information, and all requried information is passed in the request data.

For example using cURL to logon, using a certificate, and not using a userid/password

curl --cookie-jar zowe.cookie.jar --cookie zowe.cookie.jar --key ./colinpaice.key.pem  --cert ./colinpaice.pem:password   --cacert ./doczosca.pem -c - -X POST https://10.1.1.2:7554/gateway/api/v1/auth/login

Where

  • –cookie-jar … is where tokens (http tokens) are saved across invocations.
  • –key…. contains the user’s private key, used to encrypt data
  • -cert … is the public certificate sent to the server as part of the TLS handshake. It is used to decrypt ( and so validate the encrypted data sent to the server)
  • -cacert …. is the TLS certificate used to validate the TLS cerificate sent from the server
  • -X POST which http protocol to use
  • https://10.1.1.2:7554/gateway/api/v1/auth/login
    • It is http using tls ( hence the https)
    • The IP address is 10.1.1.2
    • The port is 7554
    • The application within the server is gateway/api/v1/auth/login.

The back end looks up the id from the certificate and finds the associated userid. This look up can be for a specific distinguish name, or part of the distinguished name, such as all those with o=bigbank.co.uk in the DN.

Curl has options so you can have the network traffic displayed (in clear text) so you can validate what certificates etc are being used.

I use curl to check out the backend before using Zowe.

You can specify a userid and password using the cURL options “–basic –user colin:passw0rd”.

TLS can validate the certificate, and then use the specified userid and password for authentication.

How to define an application

You can configure the back end server for Zowe and z/OSMF in different ways

So what does Zowe do?

The Zowe Command Level Interface implements the REST API and hides some of the complexity, of what headers are needed. You can provide a system wide configuration file containing the default parameters for all users, a team/project wide for the defaults specific to a team, and a person file of parameters just for you.

Zowe has been designed to work with The VSCode Interactive Development Environment, and so you can edit files on your workstation, and have them copied back to z/OS when you save the file. You can look at spool files, and issue operator commands. All this in a familiar development environment.

Leave a comment