IBM System storage DS6000 Series, Redbooks

Page: 244 / 406

220
DS6000 Series: Concepts and Architecture
11.1 What is the challenge?
In recent years we have seen an increasing speed in developing new storage servers which
can compete with the speed at which processor development introduces new processors. On
the other side, investment protection as a goal to contain Total Cost of Ownership (TCO),
dictates inventing smarter architectures that allow for growth at a component level. IBM
understood this early on, introduced its Seascape® architecture, and brought the ESS into
the marketplace in 1999 based on this architecture.
11.1.1 Speed gap between server and disk storage
Disk storage evolved over time from simple structures to a string of disk drives attached to a
disk string controller without caching capabilities. The actual disk drive—with its mechanical
movement to seek the data, rotational delays and actual transfer rates from the read/write
heads to disk buffers—created a speed gap compared to the internal speed of a server with
no mechanical speed brakes at all. Development went on to narrow this increasing speed gap
between processor memory and disk storage server with more complex structures and data
caching capabilities in the disk storage controllers. With cache hits in disk storage controller
memory, data could be read and written at channel or interface speeds between processor
memory and storage controller memory. These enhanced storage controllers, furthermore,
allowed some sharing capabilities between homogenous server platforms like S/390-based
servers. Eventually disk storage servers advanced to utilize a fully integrated architecture
based on standard building blocks as introduced by IBM with the Seascape architecture.
Over time, all components became not only bigger in capacity and faster in speed, but also
more sophisticated; for instance, using an improved caching algorithm or enhanced host
adapters to handle many processes in parallel.
11.1.2 New and enhanced functions
Parallel to this development, new functions were developed and added to the next generation
of disk storage subsystems. Some examples of new functions added over time are dual copy,
concurrent copy, and eventually various flavors of remote copy and FlashCopy. These
functions are all related to managing the data in the disk subsystems, storing the data as
quickly as possible, and retrieving the data as fast as possible. Other aspects became
increasingly important, like disaster recovery capabilities. Applications demand increasing I/O
rates and higher data rates on one hand but shorter response times on the other hand. These
conflicting goals must be solved and are the driving force to develop storage servers such as
the new DS6000 series.
With the advent of the DS6000 and its server-based structure and virtualization possibilities,
another dimension of potential functions within the storage servers is created.
These storage servers grew with respect to functionality, speed, and capacity. Parallel to their
increasing capabilities, the complexity grew as well. The art is to create systems which are
well balanced from top to bottom, and these storage servers scale very well. Figure 11-1 on
page 221 shows an abstract and simplified comparison of the basic components of a host
server and a storage server. All components at each level need to be well-balanced between
each other to provide optimum performance at a minimum cost.