RE: iSCSI: Session Partial Resolution

["Douglas Otis" <dotis@sanlight.net>]

Mike,

> -----Original Message-----
> From: Michael Krause [mailto:krause@cup.hp.com]
> Sent: Wednesday, September 20, 2000 1:36 PM
> To: Douglas Otis
> Cc: ips@ece.cmu.edu
> Subject: RE: iSCSI: Session Partial Resolution
>
>
> At 08:03 AM 9/20/00 -0700, Douglas Otis wrote:
>
> > > [D] The bandwidth (transfer rate, MB/sec) supported by storage
> > > controllers is rapidly increasing, due to several factors: (1)
> > > Increase in disk spindle and controller performance;
> >
> >The linear increase in the performance of the mechanical aspects of the
> >drive have not kept pace with the exponential data density improvements.
> >Unless dealing with a single massive transfer, the overall effect of data
> >density improvements is a small reduction in latency lost by the large
> >latency within a MAN network.
>
> Product deployment across all distances with a large percentage
> within the
> data center itself since IP / Ethernet backbone is and will remain the
> dominate technology.  As such, the latency benefits within the
> data center
> and LAN environments should not be dismissed.

If the point of splicing a WAN into the normal client<->controller<->storage
is between the controller and storage, then all residing within the same
facility will not see significant impact nor is the existing client
interfaces altered by necessity.  Placing the WAN between the client and the
controller, you have removed normal protection offered by the controller
against latency and by necessity altered the client interfaces.  Such an
architecture is at a serious disadvantage.  A WAN interface between
controller and storage also prevents a controller from being a point of
failure with respect to the remote facility.

Doug

> > > (2) Use
> > > of ever-larger caches, and improved caching algorithms;
> >
> >Network latency negates use of remote caching.
>
> Same counter argument w.r.t. latency.
>
> Mike
>