Hello:
I recently came across the LVS web site and I must say that
I am impressed and excited to begin exploring this technology.
Looking at the diagrams and reading the material has lead me
to many questions about how I can apply this to my business.
We develop JSP-based web applications that use database
back-ends.
Our current deployment uses these platforms:
Redhat Linux 7.1
Apache
Tomcat
MySQL
Each server runs software RAID for mirroring and its
own firewall.
Everything is rack-mounted in colocation spaces. The
databases are behind a double-firewall configuration.
I am thinking of using LVS for multiple
application servers with a back-end database cluster.
The reason to do this is to avoid any single
point of failure.
Here is what I have come up with so far:
I can get two 100Mbps ethernet drops from my
colocation provider. These drops
will come from separate routers to
avoid dependency on a single router.
Each of these drops will plug-into an IP switch.
Lets call these the "front line" switches.
I will have two load balancers connected to each of
the front line switches (each load balancer
has to have 2 ethernet cards).
I will have N (minimum 2) application servers connected to
each of the switches. The application servers will be
running apache and tomcat and have a separate copy of
the application software.
Since the application servers must be able to connect to
a database cluster and I want to push the database cluster
as far back from the Internet as possible, there will have
to be a "back line" of two IP switches. Each application
server will be connected to the front line switches and
the back line switches. This means that the application
servers will need 4 ethernet cards.
Behind the back line switches, there will need to be
M (minimum 2) database servers in a cluster. Each
of these machines will have 2 ethernet cards to connect
to the back line switches.
Questions:
1. Is there a better hardware configuration to use?
This design seems to be pretty intensive on
connectivity.
2. MySQL currently supports replication, but not
clustering. Is there an open source DB that
currently supports clustering?
3. I was thinking that the database could be stored on
a RAID 5 storage unit. I have no idea on how to
integrate that into the design.
I would really appreciate some help with this, since it
is bigger than I can chew right now.
Thanks,
Neil.
--
Neil Aggarwal
JAMM Consulting, Inc. (972) 612-6056, http://www.JAMMConsulting.com
Custom Internet Development Websites, Ecommerce, Java, databases
|