Actions for ERASURE CODING BASED CAUSALLY CONSISTENT SHARED MEMORY EMULATION ALGORITHM

Email RIS file
Bookmark
Report an Issue

ERASURE CODING BASED CAUSALLY CONSISTENT SHARED MEMORY EMULATION ALGORITHM

Author
Lyu, Shihang
Published
[University Park, Pennsylvania] : Pennsylvania State University, 2018.
Physical Description
1 electronic document
Additional Creators
Cadambe, Viveck
Access Online

Availability

I Want It
I Want It

Finding items...

Graduate Program
Restrictions on Access
Open Access.
Summary
This thesis develops CausalEC_exp, a causal read-write shared memory emulation algorithm for a distributed data store that uses erasure coding. In a message passing system with servers and clients, where every client is associated to a particular server to which it issues read and write operations, this thesis shows that every execution of CausalEC_exp is causally consistent.CausalEC_exp retains a key benefit of previously designed replication-based algorithms - every write operation is "local", that is, a server performs only local actions before returning to the client that issued a write operation. Furthermore, for servers that store object values in uncoded form, CausalEC_exp also ensures that read operations issued to those objects are local.This thesis shows that every operation issued to CausalEC_exp algorithm eventually terminates. In CausalEC_exp, some servers store object values corre- sponding to multiple versions temporarily, but we show that in an execution where, after a point where no new write operations are issued, the storage cost of every node eventually is equal to that of the underlying erasure coding scheme.This thesis also discusses potential benefits for using this causal shared memory algorithm that uses erasure coding, and points out directions for future work.
Other Subject(s)
Genre(s)
Dissertation Note
M.S. Pennsylvania State University 2018.
Technical Details
The full text of the dissertation is available as an Adobe Acrobat .pdf file ; Adobe Acrobat Reader required to view the file.
View MARC record | catkey: 24468673