San Francisco

dave spink toolset




SALES ENGINEERING:


OPT IO DISCOVERY STACK


OPPORTUNITY

Here is what you need to execute and achieve:

  1. Qualify the opportunity EARLY
    • Executive engagement
    • Availability of budget
    • Ability to move forward on an agreed timeline
  2. Establish the business value of the solution UPFRONT
    • TCO/ROI models
    • Operational efficiencies
    • Compelling data-driven analysis - Current State vs. Future State
  3. Establish the superiority/maturity of the solution
    • Leverage logos and references
    • Share customer success stories and business impact
  4. POC(Optional)
    • Draft a test plan and establish clear success criteria
    • Execute POC.
    • Document results, highlighting the relevant data points.
    • Present the final results to the executive engaged in Step 1.


IO DISCOVERY

Before deploying an application onto a SAN or NAS, it's helpful to know things about the application IO profile.

Application Details:

  • Is this 24 x 7 application?
  • What's the business purpose and criticality of the application?
  • Does the application have an SLAs, RTO, RPO?
  • Is the application internal or customer facing?
  • Is there an hourly, daily, weekly, monthly, or seasonal variance?
  • Is there historical performance data from which we can review and analyze?

IO Type and File System Layout:

  • How much I/O is the application doing today?
  • What is the access pattern i.e. Read / Write mix and Sequential / Random mix?
  • What is the I/O size?
  • Can the applications be tuned to write to X number of file systems and directories? Or the schema already pre-defined?
  • Can the application spread data over multiple NFS mounts points and potential multiple ZFSSA appliances?

Backup & Recover:

  • What are the requirements for backing the data and recovery?
  • Is there an archiving need?
  • Is there a replication requirement?
  • How do you backup today?
  • What RMAN backup policy is in place?
  • What are the retention policies?
  • How long does a backup take?
  • How long does a restore take?
  • What specific challenges are they having?
  • RPO/RTO?

Database:

  • DB size?
  • DB version?
  • OS version?
  • DB type i.e. data warehouse, OLTP or mixed?
  • Is the database running in archive log mode?
  • What's the daily change rate?
  • What is the month over month grow rate?
  • Do you have historical AWR performance data?
  • Total numbers of users?
  • Total number of concurrent users?
  • Is there an hourly, daily, weekly, monthly, or seasonal variance in the DB usage?
  • What is the access pattern i.e. Read / Write mix and Sequential / Random mix? (if you have AWR collections, that will tell us)
  • Is the DB planned for 12c upgrade?

Hardware:

  • Servers specs, OS, available IO slots?
  • Existing production SAN?
  • Existing production backup target?
  • Available Network bandwidth?


STACK

This table helps non technical managers understand the resources needed to address performance problems. In most cases tuning the upper layers (application, sql statements) provides more benefit then tuning the lower layers.

Host Side - Work Up User Application
Database
File System
Volume Manager
Operating System
Host Devices
HBA Drivers
Switches - Drill Down Port Utilization
InterSwitchLinks
Switch Aggregate Utilization
Storage - Drill Down Host Directors
Cache
Disk Directors
Disks