[Spread-users] Spread 4
Theo Schlossnagle
jesus at omniti.com
Sun Jul 3 10:24:12 EDT 2005
Yair Amir wrote:
> The new version has considerable changes from Spread 3, changing
> the way virtual synchrony sets are reported in membership messages.
> This is done for a good reason - to support a stronger semantics, that
> comes
> for free, for membership VS sets (instead of who-came-with-us, we can now
> know who-came-with-whom, which happens to be very useful in the
> context of
> fast synchronization as well as for some security algorithms). Of
> course, the
> [snip]
> - Ability to dynamically add and remove machines from the
> configuration, with
> deep optimization for the common cases everyone is screaming about
> for a few
> years now. We at Spread Concepts experimented with a few different
> solutions
> [snip]
> - Inclusion of the Flush layer (virtual synchrony) as an integral part
> of Spread.
> [snip]
Many congratulations on the up and coming release of Spread 4. It is
very exciting!
The who-came-with-whom change sounds nice for optimizing replication
systems with commutative semantics. It will require a lot less
application-level coordination. Very excellent.
I'd imagine the robustness being discussed isn't simply some patches
here and there. Our robustness checks (just inteneral to software, not
including byzantine network nodes) consist of automated test suites that
excercise the systems with clear code coverage assessments -- making
sure every line of code (every possible code path) is executed during
when performing the test suite. This can be a challenge for a product
like Spread (being distributed), but excellent coverage is still
possible. On the network side of things, byzantine nodes are the common
focus, but Spread 3, for example, can have some catastrophic failures
from simple node misconfiguration. While I know that a protocol
engineer would consider a single node with a bad configuration as
byzantine, a deployment engineer would not -- they'd consider it a
misconfiguration and expect it reported as such :-)
Best regards,
Theo
More information about the Spread-users
mailing list