thanks Sugu - that is helpful.
toggle quoted messageShow quoted text
If we are going to ask Kyle if he could 'do a jepsen', it would be
very good to understand which consistency choices should be tested.
On Tue, May 16, 2017 at 11:59 PM, Sugu Sougoumarane <sougou@...> wrote:
Vitess offers different transaction modes and isolation levels as described
The choice depends on what an application can tolerate. In case of the
current Vitess community, users have rarely requested 2pc. They are usually
satisfied with single-shard ACID transactions. The few that have requested
2PC are satisfied with the default isolation trade-offs.
On Tue, May 16, 2017 at 1:46 PM, Alexis Richardson <alexis@...>
This transition: X --> scalable X, suggests we are seeing parallel
evolutions of each category X. If so then we can ask what properties
of X are preserved. Atomicity and durability only?
On Tue, May 16, 2017 at 9:31 PM, Brian Grant <briangrant@...>
Again, the main value as I see it is that Vitess facilitates a
from MySQL to a more scalable implementation (scalable enough for
without taking so big a leap as NoSQL. Sugu confirmed that Vitess is
interest from MySQL users.
I could imagine also trying to find projects covering other categories,
as object storage and some flavor(s) of NoSQL.
I view filesystems as a different category than these application-level
On Tue, May 16, 2017 at 8:29 AM, Alexis Richardson via cncf-toc
The main doc:
The original presentation a few weeks ago:
cncf-toc mailing list