Connect and NonStop TBC 2024: Reflections after TBC2024

Introduction

“What do you expect from AI usage in NonStop?”  It was one of the topics that I discussed during the Beer Bust on Monday evening with a customer from the Middle East. Later, on Tuesday, Khody Khodayari from Idelji showed how their iManAlge solution uses a NonStop AI model that learns from a running application and holds the promise of automated operations. It is too much to explain here, and I am sure that Khody will do a much better job than I can, but my takeaway is this: The information that comes from Web Viewpoint Enterprise is fed into the algorithm which learns about what is “normal” and what is “abnormal” behavior. The algorithm knows how to respond to known events and learns from how “new events” are resolved. This, what I call “adapting to operational change”, is truly fascinating technology.

Just before TBC, the L24.08 RVU had been released. Below are the highlights of this RVU as the relate to NonStop SQL. These highlights are followed by some observations and thoughts that came to mind talking with other parties during the event and during my hiking trip through northern California thereafter.

NonStop SQL/MX 3.9

With SQL/MX 3.9 more cryptographic functions have been added to the DBMS_CRYPTO system package of PL/MX, the new procedural language for Oracle-like stored procedures and user-defined functions. The DBMS_CRYPTO package allows customers to create application specific encryption on columns of user tables, either SQL/MX tables or SQL/MP tables.

User defined functions can be part of SQL statements and can be executed within ESPs and disk processes if they match the criteria to do so.

OSS based MXCS using single port

A new version of the SQL/MX MX Connectivity Service (MXCS) subsystem was introduced with SQL/MX 3.9. This version has the same functionality as the current Guardian-based version but requires only one port to be opened in the firewall. All clients will use sockets on the same port as the one that is used to make the initial connection. In the new version, the OSS based association server uses AF-UNIX sockets to communicate with the mxosrvr processes and passes the socket descriptor of the client to a free mxosrvr process which creates the database connection with that client.

This version of MXCS can run side-by-side with the pre-3.9 version. There is no immediate need to migrate the existing environments to the new version. However, reducing the number of ports may be a good reason to consider the move quickly.

WebDBS and NSMF 2.0 for SQL/MX DBS

I have published several articles about SQL/MX Database Services and session TBC24.TB20 discussed how “standardization, automation and best practices are key to velocity in managing databases.” The Web Interface is now shipped with the L24.08 RVU. It uses NSHTTP as the web server and uses the REST functions for SQL/MX DBS that are included in NSMF 2.0.

Modernizing Enscribe applications

With the increased focus on security, it may be appealing to migrate data from Enscribe applications to either NonStop SQL/MP or to SQL/MX native tables.  There are two products that use intercept technology to replace Enscribe calls with calls to the SQL Engine. Escort SQL provided by comforte supports SQL/MP as the target database while E-2-S from TANDsoft offers a choice between migrating to SQL/MP or SQL/MX tables. Migrating to SQL/MX of course offers the most benefit since it allows the use of referential integrity constraints, hash-partitioned tables and a more fine-grained user access control.

If you are not ready to migrate applications to SQL, but still would like to provide SQL access to Enscribe data, there is a tool provided by Ascert that allows using an on-platform command interface or JDBC access from Java programs including the popular database-independent DBeaver GUI.

50 years of NonStop computing

While everyone came to celebrate 50 years of NonStop computing, I realized that I started my career in IT the same year (1976) as Tandem Computers shipped the first systems. I also celebrated my 35^th NonStop anniversary this September. Before joining Tandem, I worked with several operating systems: DEC System 10, Honeywell Bull OS/2000 and GCOS 6, Harris Vulcan/VOS, Harris CX/UX and Guardian/NonStop OS.

The only one left from the list is NonStop OS and for a good reason. It provides the horizontal scalability and the fault tolerance that made it famous. More importantly, however, it is the only OS to my knowledge that is not a common OS like the others: it is a cluster-operating system that manages up to 16 nodes as one single system. We in NonStop are so used to seeing this as a single system that we tend to forget that NonStop OS supports already what “the cloud” is still not capable to provide, because they are using an OS (Linux) that manages only a single node.

The latest application architectures that are based on Kubernetes (K8S) require (multiple) separate orchestrators to manage a herd of container instances. In September our partner Lusis and HPE showed how NonStop could achieve the same elasticity for its Tango application without using K8S but using Pathmon and NonStop TS/MP instead.

Lusis concluded in a LinkedIn article at https://www.linkedin.com/pulse/lusis-tango-elasticity-hpe-nonstop-philippe-pr%C3%A9val-lusis-qi5oe/

The idea is to use “NonStop TS/MP Pathmon” and to give it the role of K8s. Tango must communicate its metrics to TS/MP and TS/MP will start new instances (including on new nodes) in case of resources struggling. TS/MP is NonStop’s service orchestrator. It works out of the box. It is simple, robust and fault tolerant.

Not bad for a 50-year-old architecture! We only need to rename our components, Pathmon and CPUs to Orchestrator and Nodes.  That being too simple, there is a need to describe the NonStop architecture in language that today’s IT professionals can relate to.