Rocket U2 talks about UniData 8.1
We sat down with Vinnie Smith, Managing Director R&D, Rocket U2, John Nunziato, Product Manager, Rocket MultiValue Databases, Michael Byrne, Rocket MultiValue Product Evangelist, John Zagnoli, Rocket U2 Development and Engineering Manager, and Heather Smiles, Product Marketing Manager, DBMS and Application Servers, to talk about UniData.
IS: Spreadsheets and Word Processors reached a point where the supposed added benefit of each release was less and less useful. Tell us some powerful, practical things that companies will get out of the new UniData release.
Vinnie: 8.1 is the biggest release in a decade. There's a number of key areas of benefit. The 64 bit architecture is more stable.
John N: The obvious effect is on the file access size; being able to address much larger files. We can now exceed the two gigabyte boundary. That's brought along reliability improvements. The process architecture now can access a 64 bit memory model. You can process much larger datasets in memory.
Michael: Account-based licensing allows administrators to reserve a certain amount of licenses for each virtual group. That way you always have enough licenses for a specific group, like admins. This really crosses over into our hosted solutions, especially anyone who is doing SaaS (Software as a Service). And it's in a configuration file. You can configure accounts into virtual groups. Everything not in the file is in the default pool. If you sell cloud services, you can control the split of the licenses between clients.
Vinnie: This will help in building a multi-tenant solution.
John N: Security. SHA1 and SHA2 security on files. We also have improvements to protect against Heartbleed and Poodle attacks.
John Z: Automatic data encryption at rest includes the keys and indexes. You can 100% encrypt everything in the data file. If you store data in your dictionary, you can encrypt that, too.
Michael: That protects you in the case of your disk drives being stolen, or your data backups.
Vinnie: And supporting IPv6. Because, as the world runs out of Internet addresses...
IS: We hear "IPv6" a lot, but I'm not sure everyone understands what it is. Can you explain it briefly?
John N: IPv6 uses 128 bit, that gives us 3.4x1038 IP addresses. [Editor's note: IPv4, by comparison, tops out at roughly 10 digits (decimal) worth of addresses vs. IPv6's 32 digits (hexadecimal) worth of addresses.] More security, including the authentication header protocol and the simplified packet header structure, which makes data packets faster and more efficient. More data security — encryption capabilities — when communicating between systems. We now offer two BASIC commands: SET-IPV and GET-IPV to enable and disable IPv6. Or allow both IPv4 and IPv6, prioritizing them. There will also be config file options.
Vinnie: Basically a dual stack capability incorporating both protocols. Calling out to other IPv6 devices — the Internet of Things — being able to call out to them is very important in terms of extending what you can do with UniData.
IS: The big fear in any upgrade is stability. Can you tell us a little about your testing process?
John Z: This was a major code upgrade, especially the 64 bit extensions. We had cross-function testing, QA testing, Support Department testing, and beta testing. Beta was August and ran through to January. After that, we had several customers run their existing applications and found no regression errors. We got acceptance from everyone in the beta.
IS: How does the upgrade work? Has the process been improved?
Vinnie: Especially in multi-system sites, upgrades can be quite challenging. We've automated out most of the manual steps. You can even add your own post-upgrade scripting. We've done our best to make the upgrades easier for everyone. We are quite happy that we've smoothed the upgrades for everybody.
IS: I know that when I work on a large, complex software project, some aspect of it becomes a pet project, something that I'm particularly proud of at the end. Do any of you have a pet feature in UniData 8.1?
John Z: I can name two. I'll start with the local subroutine. If you have a large mainline program and you want to add a subroutine, you can now build it into the body of the program using variable scoping. Everything compiles as a single unit. We had already released it in UniVerse 11.2. We extended it into UniData as well. It's an internal subroutine but it has separate variables like an external subroutine.
Michael: It helps with repetitive work where you might otherwise risk stepping on global variables.
John Z: The other is the replication performance monitor. You can see each file and how it is changing: data utilization, buffering... This is at a level you haven't been able to see before in any other monitoring tools. (See video interview with Jonathan Smith at: https://www.youtube.com/watch?v=t6NzEd2XXio)
Michael: Speaking of pet projects, Python isn't in this release, but 8.1 gives us the groundwork. We are laying the foundation for Python while thinking about how we can extend that architecture into other languages. I find that exciting to me; getting connected to the outside world. Making our databases available to a whole new audience of developers who know other languages.
IS: UniData, UniVerse, and D3 are all part of the Rocket family now. Is there a feature or aspect that sets UniData apart?
Michael: In my opinion, from a reliability standpoint, UniData has always had RFS and Replication in the engine. We have several financial — big customers — that depend on that.
The database servers are all somewhat similar, especially as we cross-implement features to make them all more robust. We're really thinking forward, as we develop for one engine... thinking of how it will be implemented in the other databases.
Heather: We also have mvBase, too.
Vinnie: We think in themes, like security, compliance, reliability. We try to love all our children equally, so to speak. A lot of our code is moving to a common code base in the middle tier — client connect — aspect. On the Rocket website, there are white papers about 64 bit and other features. We also have lots of examples on Github. You can also follow us on LinkedIn, Twitter, and Facebook.
IS: Anything "deeper-dive tech" that you wanted to talk about in this interview?
John Z: Dynamic files. Not only are our dynamic files 64 bit, but we've added a new hashing algorithm. We've also improved the merge and split process. It has been tested against extremely large files.
John N: For replication, we added data compression — LZ4 — to reduce the footprint between publisher and subscriber by at least 50%.
IS: Are there any other new UniData features that you'd like to talk about?
Vinnie: We covered the high-level list. I do want to point out that we have an upcoming MV University. We'll be discussing many of the 8.1 features in more detail.
Heather: We've already sent out a save the date for Harrah's in Vegas.
Charles Barouch is the CTO of HDWP, Inc. He is also a regular contributor to International Spectrum Magazine, a former Associate Editor for both Database Trends and for Gateways Magazine, a former distance learning Instructor for CALC. He is presently the Past President of the U2UG. Mr. Barouch has presented technology and business topics in front of hundreds of companies, in a wide range of product and service categories. He is available for on-site speaking and consulting engagements in and out of the United States.
View more articles