Which optimization approach should be implemented?

[zsStrike]

In the chapter 3.8.4 in K23SI-Detailed-States document, there are two approaches for general optimization:

• the client keeps track of what write operation has already got persistence confirmed...
• the TRH keeps track of what write operation has already got persistence confirmed...

I wonder if the two approaches should be both implemented or just implement the latter as we discussed in last few weeks?

[iafuture]

i believe this was answered in the meetings. The optimization would be to allow TRH to keep track of completed writes