deletions | additions       diff --git a/Questions.md b/Questions.md  index e0d114f..0936ec3 100644  --- a/Questions.md  +++ b/Questions.md 
...
- in Spark?  - in an existing database?  - Redis? ("data structures server", but not designed to be consistent across replicas)  - would people take this seriously if it was a modification to Redis?  - Tapir?  - Doppel? (Phase Reconciliation paper)  - code not (yet) available  - Lynx / Rococo  - Distributed key/value store  - Existing DTM framework  - HyFlow (Scala)  - how best to evaluate?  diff --git a/Related work.md b/Related work.md  index 7a3f42a..4a7bc9e 100644  --- a/Related work.md  +++ b/Related work.md 
...
- doesn't do replication for hot keys  - how pieces are reordered seems a bit crude (anything that produces a value is "immediate" and cannot be deferred)  - seems like it would have high runtime overhead (tracking dependences dynamically, despite using offline (static?) analysis)  - Citations  - ###  Transaction chains - Prior work  \cite{Zhang:SOSP13} (same authors): *Lynx*; statically split authors)  - *Lynx*: geo-replicated key/value store  - programmer splits  transactions into pieces that run on different locations, either the locations  -  whole chain gets committed or none is executed in a serializable way, can only *fail* at the first hop  - static analysis to determine if the above can be enforced  - *Benchmarks*  - Social: FB-like social graph, users, walls, activity feeds  - Twitter: Users, tweets, directed graph  - RUBiS  - *Questions*  - not thinking about high contention cases? does that come up in the social graphs?  ## Enhancing Concurrency in DTM through Commutativity   - \cite{Kim:EuroPar13}: EuroPar'13, Junwhan Kim, Roberto Palmieri, Binoy Ravindran