Paper evaluations

Kiran Tati (ktati@csag.ucsd.edu)
Thu, 27 Apr 2000 01:01:28 -0700

Experience with Grapevine: The Growth of a Distributed System
The primary objective of this paper is to prove the design decisions
are correct and how the Grapevine is scaled well with the time. The
Grapevine is a simple and rudimentary messaging system. It has two
servers namely message and registration services. The message server is
used mainly to move the data around in the network and registration
server is used to maintain the information about the users
(authentication, access control) and resource location. This system
achieved reliability by carefully replicating data maintained by the
registration server. This replication and distribution is transparent
from the user by proving the library functions with occasional surprises
(User some time can't able to the newly added user instantly because of
distributed and it is not transparent for the Regis raters). This system
is scaled well from the 2500 messages/day to 8500 messages/day. It is
quite interesting to see the same problems that existed today with
e-mail system (Spam mails). This system is well cope up with the load
but some of decision are changed in order in corporate the load
(shortening the update messages by sending only changed information
rather than sending entire registry). The special mechanisms (logs and
remote access windows) are provided to operate remotely for debugging.

Fine Grained Mobility in the Emerald System:
The main goal of this paper is designing a distributed object system
that support object migration. The important design decision is that
proving global naming for all objects, which potentially moves, in fact
this is absolutely required. This system has an advantage over the
traditional operating system in which the operating system has some
state associated with the each process, which is hard to transfer from
one node to another node (especially pointer to the data structures
maintained by the operating system). This language introduces a new
parameter passing mechanism, call-by-move and call-by-visit, which is
very interesting. I think it is better to use the call-by-value
semantics rather than supporting the call-by-reference mechanism for
parameter passing. The location of the object is found by using the
forwarding mechanism. It is not clear in this paper about how the system
handles the in transit objects.
The design is favored improving the local call at the expense of
overhead with the moving objects. The objects are referenced using
direct pointer rather than indirect pointer to improve the performance
local call invocation. This really complicates the migration of object.
There are two problems in moving the object that are specific to this
system are location activation records and object references stored in
the registers. A simple garbage collector is provided to clean up the
resources used by the objects.

--
Kiran Tati (Graduate Student)            Kiran Tati
UCSD/CSE AP&M 4402                       9450 Gilman Drive
9500, Gilman Drive, Dept. 0114           UCSD#922563
La Jolla, CA, 92093-0114                 La Jolla, CA 92092-2563
Phone:  858 5345486 (Office)             858 552 9291 (Home)
E-Mail: ktati@csag.ucsd.edu


Office Address: AP & M 6426