Virtualization and Cloud executives share their predictions for 2015. Read them in this VMblog.com series exclusive.
Contributed article by Dr. Michael Waclawiczek, Vice President, Operations, NuoDB, Inc.
2015: Suddenly It's Cool To Scale SQL
So it's nearly 2015 and almost all aspects of the
application and infrastructure stack can be virtualized and scaled to achieve
higher throughput and high availability requirements. This is also true for the database layer as
long as all data can be stored and managed by a single server or virtual
machine. If you are using a traditional
relational database as part of the stack, scaling up on a single machine is
pretty much the rule unless you are willing to break your database into shards
or replicate data across multiple machines.
The relational DBMS vendors have spent years perfecting
workarounds to overcome these inherent limitations in the original architecture
of their products. These systems where
designed for the client/server systems back in the 1980s, not for modern datacenters
or the cloud both of which embrace a scale-out rather than a scale-up application
deployment model.
Using these workarounds to deploy multiple client/server
databases as a single, logical, distributed scale-out system work and are
proven in production. These systems have been carefully designed and often deployed
at enormous expense. They work as long as they don't hit the next wall from a
growing user base, database growth, or other scaling challenges.
New approaches to turning SQL databases into truly
distributed systems are fundamentally different from the original relational
database designs. They look like relational databases from the outside in but
from the inside out they are innovative, new architectures that resemble some
of the approaches taken by in-memory databases and NoSQL.
2015 is the year that NewSQL database systems are crossing
the chasm and will begin supporting true enterprise-scale applications previously
based on the big name RDBMS systems.
2015 will also mark the start of a major shift in how
companies and application vendors evaluate migrating business-critical on-premise
applications to cloud-based architectures.
The fear that cloud-based deployments of relational database
applications won't scale will end. Other
fear factors like data security are perennial and won't cease despite the
ability of the new-age distributed databases to provide the same levels of
security and encryption mechanisms found in mainstream RDBMS systems today.
NoSQL systems have paved the way for getting distributed
systems into the mainstream. However
NoSQL necessitates trade offs between transactional consistency and
scalability. These trade offs are
acceptable for various operational and analytical applications that don't require
the ACID compliance offered by RDBMS.
NewSQL systems will pave the way for scale-out performance with
SQL. Watch out for big news in this space in 2015.
##
About the Author
Dr. Michael Waclawiczek is a highly respected marketing and
product management executive in the enterprise software industry with over 29
years' experience. He has launched more than 20 major software products
generating over $1B in total revenue. Michael has held several executive and
senior management positions at private and public companies including Expressor
Software, StreamBase Systems, Kalido, IONA Technologies, Object Design, ICAD,
and Unigraphics. Prior to his corporate experience,
Michael was Assistant Professor at the Technical University of Vienna, Austria,
where he also received his Master's and Doctorate degrees in Mechanical
Engineering.