You do not need to send archive data immediately to some far off, unreachable storage facility. In fact, you can archive from one Adabas file to another Adabas file. You do not have to go directly to an off-host storage facility. You can archive out to Adabas on another (less expensive) architecture for example. This means the data is in the same form and can be reached by your existing applications.
If you archive to another Adabas file you might use that file as an intermediate storage facility (for the most recently archived content) and then still archive from that intermediate file to a flat-file store at a later date. According to the data life cycle suggested earlier, you can have your present-day data in primary systems, your recent data in secondary (Adabas) systems and your historical data in a real flat-file store.
Obviously you do not need to use the intermediate Adabas file approach, you can go direct to a flat store. Conversely, you may only wish to archive to an Adabas store. It is your choice, on a file by file basis.
It has long been known that Adabas Vista partitioning can be used to segregate data by criteria that is relevant to your business. In many cases this is in some way related to the data life cycle. For example, partitions may be organized by date. This is clearly a situation where the older partitions can be a) created by Data Archiving for Adabas as the age increases and/or b) by actually archiving away the very oldest data out of the Vista partitions altogether. At a later point it may be that older data is recalled into a temporary partition to allow old data to be included into your production systems as you need. This means you can use your existing systems to process data recalled from your archive at any time.
You might also recall archived data into a standard Adabas file (not partitioned) but use your normal applications but with a Vista translation rule to redirect activity to the recalled file rather than to the actual production data. This is another way you can use your existing systems to process data recalled from your archive at any time.
Once data is archived to a flat store you do not have to worry. It remains within easy reach. You can search it without having to recall it to Adabas. Obviously, you can perform full Adabas searches if the archive is an Adabas file. You can search the flat store too. Clearly, searching a flat store is not as flexible or fast as when searching with Adabas. However, the flat-file search is only needed to get you to the point where you identify the data you need. Then you recall this data into its original (Adabas) form, where once again you have full Adabas search capability.
You can recall whole archive runs that were performed previously or you can recall the results of searches performed on the archive, it is up to you. You can recall into an empty Adabas file or into an existing Adabas file, provided the original FDT is supported by the destination Adabas file.
Let’s say we archive 1 million records per month to a flat-file store. After one year we have 12 million records. After ten years we have 120 million records. This example spans several years where there are (12*10=120) different operations to put information into the archive. In ten years, how many times does the data structure change? “Many times” is the likely answer.
Data Archiving for Adabas takes the meta-data (FDT) along with the data and manages change to data structures. If you wish to recall data to a new FDT that does not support all the fields that were included in the original archived data you can choose to ignore the extra data, for example. Recognition and tolerance of these changes over long periods of time is very important in the tooling you use.
You can choose to transfer data to another file instead of archiving to the vault. In doing so you can choose to move the data or just copy it, your choice. This can be performed adhoc or to a regular schedule. For example, you might set a regular schedule so that every month 25% of a production file is transferred (copied) to another Adabas file for testing purposes (and obviously run a program to disguise the data from the original production values for privacy, legal reasons).
Similarly, you can decide to use Data Archiving for Adabas to regularly remove data that is not to be sent to the archive because it serves no further purpose, for example, the expiration date has been reached.