If you are planning to use a PDS class, you must consider whether individual users of your class are likely to have private data sets concatenated ahead of eligible data sets (data sets that contain members that VLF is to manage as data objects). VLF is less efficient when private data sets are present in a user's concatenation ahead of the eligible data sets.
Your application does not have to distinguish between private data sets and eligible data sets. VLF, however, does perform extra processing when private data sets precede eligible data sets. Specifically, if an end user has a private data set concatenated ahead of an eligible data set, the first attempt to retrieve a given object from VLF storage will not succeed.
If you have control over the search orders, VLF works most efficiently when private data sets (or ineligible major names for non-PDS classes) are either not allowed or follow the eligible names rather than precede them.