In reading through the recent OMB status report, I keep seeing references to systems having been "implemented". The report states that the gov deadline to "implement" systems is March 31 and that tier 1 agencies are 96% complete, tier 2 are 74%, etc.

By "implemented", does this mean that this percentage of systems have been tested, put back into production and are currently being used? Based on previous reports, I find this very hard to believe. Given the state of their data exchanges alone, I would think things would be blowing up left and right.

Can someone shed some light?

Thanks!

R.

-- Roland (nottelling@nowhere.com), March 19, 1999

Answers

Roland,

Remember that if a remediated system has been "implemented," that simply means that it has been put back into production today, where it will continue to run with today's vintage-1999 dates for another 287 days. The implementation/deployment decision does help ensure that the act of remediation has not introduced any other extraneous logic errors (sometimes known as the "bad fix" phenomenon), but it doesn't really tell us whether the system will run properly with post-2000 dates -- unless, of course, it's a system that, by its very nature, carries out "look-ahead" calculations.

In fact, even the statement that a system has been "tested" doesn't necessarily tell us much, because we need to know whether it's been tested with vintage-1999 dates or vintage-2000 dates. To do the latter (which is, of course, what we would prefer to see everyone doing) usually requires us to "age" the database that's being used as part of the test environment, so that all of the pre-2000 dates are rolled forward in the same fashion as the hardware clock is rolled forward on the CPU in order to make the programs operate in a post-2000 environment. Most organizations find that it's simply too time- consuming and difficult to do the database-aging, though there are some clever ways (so-called "dynamic aging") of doing it...

As always, none of these statements mean a thing unless they've been verified and validated by an independent third-party reviewer ...

Ed

-- Ed Yourdon (ed@yourdon.com), March 19, 1999.

Ed -

Thanks for your thoughts...I quite agree with your analysis. The question outstanding in my mind relates to the passing of dates and date sensitive data between systems. Unless the interfacing systems are compliant, major problems can occur, as we all know.

Given the number of interfaces that surely exist within these agencies' systems, I find it nearly impossible to believe that they have put 76% of their "compliant" systems back into production. Yet if they have not, the report is totally worthless and misleading, which would come as a huge shock to me.

R.

-- Roland (nottelling@nowhere.com), March 19, 1999.

Roland:

The situation you describe is one of the main arguments for using windowing techniques.

When windowing is used, the date formats in the interface files do not have to change, allowing systems to be implemented back into production interfacing with other systems that may or may not have finished remediation.

-- Hoffmeister (hoff_meister@my-dejanews.com), March 19, 1999.

I realize this...but I have also read that the government is primarily using expansion to solve the problem. Which leaves my question unanswered...

R.

-- Roland (nottelling@nowhere.com), March 19, 1999.

Primarily using expansion? What is your source of this? This recent article seems to indicate a mixed bag of windowing and expansion in the government:

hoff_meister@my-dejanews.com), March 19, 1999.

Hoff -

I cannot find the article...was in a local paper and could very well have been wrong (no big surprise here). And yes, the approach is a mixed bag.

But consider that SSA (which I cannot see utilizing windowing), NASA, FEMA, Education, OPM, HUD, Interior, GSA, VA, SBA, EPA, NSF and NRC have reportedly implemented 96% of their systems. This means live, in production in my book.

My only question is, is this FEASIBLE and if they are in production, how in the hell are all the interfaces functional??

R.

-- Roland (nottelling@nowhere.com), March 19, 1999.

You have to realize that, even with an application that was remediated using expansion, interfaces to/from that system may and probably will in some instances still use windowing.

-- Hoffmeister (hoff_meister@my-dejanews.com), March 19, 1999.

If windowing - what is the "government standard" definition for which window to use? yes - many are using both, in mixed ranges as I understand form reading the government computer publicly available news.

A mixed bag from different departments using different methods internally, between divisions in each department, and between the various departments, and between agencies too.

Most seems to depend on when the specifc repair effort began. Data exchange between any two or threee computers seems to be: leave the file itself alone - very little "exchanged" files appear to be expanded to 4 digits, most stay the same because the "otehr side" may or may not be ready to receive new dates, handle new dates, or export its own dates to anything else. Hence, a remediated process at one place gives 2 digit dates in output to something else, which may or may not be fixed yet. So continued output in "XX" form is the normal response.

But the assumptions of windowing must occur when data is read in, and when it is written out: is "45" 1945, or 2045. For my grandmother, is "98" 1898 or 1998? Or 2098?

Expect many and varied problems, I just hope that they try to enforce the stupidity of computers with the stubborness of bureacrats.

-- Robert A. Cook, P.E. (Kennesaw, GA) (cook.r@csaatl.com), March 19, 1999.

My last post on this issue.

That is assuming the systems they are interfacing to are compliant, albiet using windowing. What if they aren't? Shouldn't we see problems cropping up NOW if those agencies have implemented 96% of their systems and the interfacing systems are not yet compliant?

I am not trying to be difficult, merely trying to understand.

Yours in technical ignorance,

R.

-- Roland (nottelling@nowhere.com), March 19, 1999.

Roland:

In general, yes, some of the "bad fix" errors would start showing up. This is a Good Thing, in that they can be addressed now, and not during a crunch at rollover.

But to your other point. The point I am trying to make is that interface files may still use windowed, 2-digit year dates, even with systems that may have used date expansion internally. At this point, passing this data to a system not yet compliant would not cause problems, since the receiving system would be receiving dates in 99 or less, and process them as it always has.

Robert:

I truly think the windowing problem is somewhat overblown. Pivot years are selected based on the data. Obviously, some dates, such as birthdates, do not lend themselves to windowing. But the vast majority do.

This is indeed a somewhat temporary fix, to get over the immediate danger of the rollover. It is also the reason I see demand for ERP software, such as SAP, to dramatically increase following the Y2k hump. Companies will continue to replace their older systems, because of the temporary nature of many of the remediation fixes.

-- Hoffmeister (hoff_meister@my-dejanews.com), March 19, 1999.

Hoff -

Well I sincerely hope that the "50" that means "1950" on my system doesn't mean "2050" on yours.

R.

-- Roland (nottelling@nowhere.com), March 19, 1999.

Roland:

As I've tried to explain, the vast majority of dates passed in interface files (from experience, a large number above 99%), are current dates, that lend themselves very well to windowing. Invoice dates, delivery dates, posting dates, etc.

Obviously, some dates do not. The use of windowing is based on the dates passed. If a year of '50' is possible to be passed, and open to misinterpretation, it is again obvious that windowing is not a viable solution.

-- Hoffmeister (hoff_meister@my-dejanews.com), March 19, 1999.