What do you mean the "log structured data store"? Does it mean K/V data will be
directly converted into blobs which are stored in blobstore?
I am very interested in what you mentioned, I wonder whether it is related to the blobkv I
have proposed?  Can you share more information with me? If you have some spare time,
please visit them. I really hope to hear more about it from you.
On 04/18/2018 09:36，松本周平 / MATSUMOTO，SHUUHEI<shuhei.matsumoto.xt(a)hitachi.com>
Thanks for you reply.
In the meeting, I got good understanding about blobstore and related modules and
and I shared that a few ideas were mainly about customized bdev modules.
My point is to develop a sample implementation suitable for SPDK of more flash friendly
(sequential write) block store with some additional features, e.g., compression.
Utilizing both front-end and back-end of SPDK and implementing customized bdev layer will
require deep understanding of SPDK to us.
Any sample implementation may be helpful not only for us but also for community.
Activity from now on will be very experimental and not organizational yet. Hence pace may
be very slow.
The first activity may be a patch with in a couple of months about indexed allocation,
direct blocks, single indirect, double indirect, ...
I hope these make sense to you and we will be able to have more discussion in near
差出人: SPDK <spdk-bounces(a)lists.01.org> が Luse, Paul E <paul.e.luse(a)intel.com>
送信日時: 2018年4月17日 23:31:10
宛先: Storage Performance Development Kit
件名: [!][SPDK] Topic from last week's community meeting
I was out of town last week and missed the meeting but saw on Trello you had the topic
“a few idea: log structured data store , data store with compression, and metadata
replication of Blobstore”
Which I’d be pretty interested in working on with you or at least hearing more about it.
When you get a chance, no hurry, can you please expand a little on how the conversation
went and what you’re looking at specifically?