Comments within. Again, I am a U2 fan, so I am trying to do a comparison, not a sales job, although I'm guessing it might sound a bit like Cache evangelism. I am pragmatic about my choices, so even if I want to like a more pure MultiValue platform (I started my career on a Pr1me computer, prior to Information being on it) over a MultiValue platform running within a MUMPS p-machine, using different underlying data structures, for example, I don't make my decisions based on any such "religious convictions" regarding tools.
That said, to the extent that there is an "MV evangelist" in me, I also like the toolset I'm using, so I'm sure I am inadvertently injecting some "spin." Please treat accordingly. U2 is a good toolset and will get even better, I have no doubt. If I had an application already running successfully on this platform, I would not jump ship without a solid business reason to do so. On Sat, Apr 9, 2011 at 5:57 PM, David Jordan <[email protected]> wrote: > >From my observations, I don't quite agree on Tony's summation of Cache vs > U2. Cache has the same burden as U2 in that they came from mumps where U2 > came from Pick. The biggest difference between Cache and U2 is marketing > where Cache takes on RDBMS and U2 is apologetic for being Pick, but that is > not the issue with the technology. This might have more to do with the advances made once they bought out almost all of the MUMPS world, where the PICK world is still fragmented. Once they had all of MUMPS, they could take all of the products and pour them into one, then run with that puppy into the future, competing outside of the MUMPS world rather than within it. That is how they are playing in the DBMS sandbox in a bigger way than each of the individual smaller MV shops. > Cache is a multi dimensional platform, the same as Pick but in a slightly > different way. U2 has SQL just as much as Cache, This is decidedly NOT the case. I have worked with UniData SQL extensively and UniVerse SQL to a tiny extent. Cache' SQL is far, far better in my opinion. In fact--and this might strike you as a bad thing but I think it is very, very good--their MV Query language executes by first generating SQL and then executing it. It is fast, fast, fast because of (and in spite of) this. You will not find a faster SQL engine against MV data. My understanding is that others have done benchmarks and Cache' has come out ahead of other MV vendor products. Additionally, the approach to the metadata is comparatively excellent. We can source all of our file metadata in "persistent classes." One button to compile generates the DICT, the SQL projection, the Model for an AJAX Model-View-Controller approach, the object metadata, XML, and whatever else anyone might need, I suspect (but these are all I have used). The "class doc" is also generated at that time, which is also terrific. We can document our metadata as handily as any other code. > I have applications written in U2 that have SQL tables and SQL grants and > can be accessed with SQL queries. If Cache can sell itself as an SQL > database, then so can U2 they are both in the same boat. Not nearly as easy a sell for U2 as for Cache. Really. Trust me on this one, David ;-) U2 would not want to put their SQL offerings up against those of Cache'. > Whilst I think some of Cache's technology is clever in their object model, > people need to consider some issues with Cache. Check that the ATOMic > transaction model is what you expect. U2 complies to the standards, when I > looked at Cache, I don't believe that they did. Interesting. I'll have to look into that at some point (or you can clue me in), but in the mean time we have seen no issues in this regard. We do get referential integrity where we want it when we specify a field as a foreign key. It's pretty cool. We can then use the arrow notation in SQL so we can skip the TRANS I-descs and do SQL statements such as SELECT AddressId->City from Party.Person; In this case, AddressId is a field in the Party.Person DICT and City is in the Party.Address file that is keyed by an autoincrement key that the AddressId points to. We even have a file template that is a superclass of each of our persistent classes which has come in very handy. > Also consider the continuity of the company, what is the future of > Intersystems when the founder exits, where Rocket is a more established > business that is not so dependent on key managers. > I agree that is worth consideration. Take a look at the financials to the extent you can, however. > > I am currently looking at how U2 fits in the cloud environment with > products like Microsoft Azure and I think the model of U2 where each table > is a separate os file is better for cloud computing than Cache's one system > file (similar to the approach of other RDBMs). > I think you have a case here, although the high availability, high scalability of Cache' might help in that regard. I do not have a clear understanding of possible shredding in either environment. FWIW, Cache' is introducing an open source cloud database that is very lean. They have the Java interface right now with the .NET framework coming, as I understand it. The data looks like MUMPS (a tree) rather than like PICK but I look forward to seeing what happens with a port of data there. That will not be high on my priority list for some time, so I'll be watching to see what others do with the open source product. > > DataVu provides a significant solution to an area of weakness in the U2 > camp related to reporting. DataVu gives U2 the equivalent of SQL Reporting > services and more. DataVu provides a competitive position to products like > Cognos. This is one technology that I do not believe that Cache can cover > off. > I do not know DataVu. Cache' has a BI tool but their reporting tool is for developers, not directly for end-users. MITS can be used for reporting too. > > I think it is great to have Cache in the MV world, it provides competition > and will keep Rocket on its toes, both have strengths and weaknesses. Yes, I completely agree. > However I would not agree in Tony's email that it is a better technology > than U2 I did an assessment at the start of my project for my purposes. If I had come down to "6 of one, half dozen of the other," I would have stuck with U2. I did not have to do a conversion. I do think Cache' is a technical platform that has advanced beyond any of the other MV vendor offerings on many fronts. There are still pros and cons, no doubt, but I feel confident I made the right decision. > or that it is a solution to the management concerns who are more familiar > with Oracle and SQL Server. Agreed. If there are still managers who demand a particular DBMS for an application, then they are unlikely to demand Cache' any more than demanding U2. However, if they have a feature set they need, it is more likely they will be able to handle the features out of the box with Cache than with any of the other MV platforms. > I also don't believe that U2 is locked to green screen applications. I am > seeing new generations of applications that don't have green screens, they > have a range of .Net, java and web interfaces. Agreed. > Honestly the PICK issue does not come up as much as people think it does, > nor is it a hindrance. Management are looking for ROI, they want responses > to business problems and they want to ensure that Risk management is ticked > off. All these can be achieved with U2 as a community we need to learn how > to answer those concerns. > > There are some areas of U2 that need to be refined and as U2 users we need > to tell Rocket where to put the priority to ensure that meets our > requirements. The U2UG has set up the better and better site and Rocket > has the [email protected] email. Please take advantage of these to refine U2 > to meet your expectations. > Agreed. If Rocket has the financial resources to do it, I think the U2 team has what it takes to move U2 to better complete with Cache'. My hunch is that InterSystems has far more resources, but I would love to see U2 be more competitive in its offerings, so I agree with you, David. Ask for those things that are most important or the biggest bang for the buck and the U2UG can be another input, like the large U2 customers are, for helping Rocket U2 choose good features for moving forward. I just noticed Wol's comment re jBASE below and will suggest that those who wrote the MV implementation into the Cache' engine would say that the spec was more UV than jBASE. Just my two cents in case it is helpful to anyone. cheers! --dawn > > Regards > > David Jordan > U2UG VP > > _______________________________________________ > U2-Users mailing list > [email protected] > http://listserver.u2ug.org/mailman/listinfo/u2-users > -- Dawn M. Wolthuis Take and give some delight today _______________________________________________ U2-Users mailing list [email protected] http://listserver.u2ug.org/mailman/listinfo/u2-users
