A couple of years ago, I was consulting for a large Canadian Telco customer. They told me they were achieving 80TB-per-person for their storage infrastructure. I found this ratio impossible to believe. As we got deeper into the topic, I found that many of what I considered basic labor functions were not factored into that number. Nevertheless, they were very proud of what they had achieved in a stead-state storage operations function.

Now, 80TB per person is not that unusual. I had a customer a month ago show me how they were measuring 250-300TB per person. I have also met with IT executives who tout the best practices and solid-management style of their teams, but only measure 30-40TB per person. These ratios/numbers are all over the board, and we must conclude that simple labor comparisons are not-so-simple.

This question of potential TB-per-person comes up frequently, as it should. Labor is roughly 40% of storage TCO, so improving labor effectiveness will reduce ownership costs. Investments in automation, new storage architectures, and best practices (like ITIL) can be shown to have a high return on investment and lower the labor cost of storage ownership. In fact there are many investments that can enable today’s staff to manage more storage, more effectively, now and in the near future.

Hu Yoshida has blogged on this topic; I have had a few things to say as well.

If you want a method or approach to calculate TB-per-person, be prepared to consider a broad array of people that are involved (directly and indirectly) with storage management, operation and cost-of-ownership. The several dimensions of the storage ecosystem that need to be considered when making these ratio decisions are outlined below:

Organizational

• Does the staff ratio include server sys admin, and their role with local storage

• How about staff that handle backup and recovery functions

• The staff responsible for DR testing, provisioning and certification

• Does the ratio include procurement staff

• Change control, version control people are often left out

• There can be a small army of NOC troubleshooting staff that are often ignored

• How about vendor consultants or SE that are on-site for tech support

Engineering Functions

• Storage architectures just don’t show up, there is engineering support for various functions of the storage lifecycle

• Certification, investigation and ramp-up time/effort for new systems

• Does the SAN support and engineering fall into the storage ratio

• How is the lifecycle of the data and its long-term usage patterns factor into the storage engineering and architecture

Operation Functions

• Operations staff are most commonly factored in the TB/Person ratio

• Relative position of storage management maturity (see the Gartner Services Maturity model to rank where you may be)

• There are many tasks within operations, see how many of these tasks your people undertake

o Search/enter data for devices, ports, & WWN

o Map config of SQL, Oracle, & file servers

o ID out-of-date fabric firmware

o ID SAN configuration changes

o ID specific devices to plan for downtime

o Correlate events to applications

o Track Storage Capacity, free space

o Usage Statistics for Capacity Planning

o Locate stranded capacity from hosts

o Policies for disk; ID unused / stale data

o Balancing capacity, access to hosts

o Isolating performance problems

o Performance stats for trending & forecast

o Provisioning new/additional storage

o Tracking assets for inventory

o Accounting for Chargeback o Creating audit trails/reports

o Time for storage/server consolidation plans

o Hours Recovering from Disasters

o Training / using storage management tools

IT Management Overhead

• There can be line managers and IT managers that are dedicated to storage, or at a minimum have responsibility for storage plans

• Storage architecture is a function that can be a dedicated function for large IT shops, or shared within a single architect for smaller ones

• Every management team has admin support (secretaries), are they counted

Business Operations

• Your type of business may require more or less data and storage management based on

o Compliance regulations and laws

o Data retention policies o Audit support and litigation support

o Disaster protection

o Recovery time and recovery points

• Your business applications will also determine if the data being storage is structured, unstructured, or semi-structured. Different data types will impact provisioning, LUN size, RAID protection, backups, migration, snap copies etc.

Storage Architecture

• The basic storage architecture has a huge impact on labor ratios and effort needed

• Is the architecture centralized, or spread across multiple sites, in multiple locations/countries/time zones

• Is disaster recovery an integral part of the architecture

• Connection types, FC, FICON, ESCON, NAS, iSCSI also add to or alter the support time

• How much of the data is server (DAS) attached

• How much is in storage silos

• What storage is SAN connected, direct attached, NAS

• What appliances or specialty servers are used for virtualization, de-duplication, backups

• Is disk based backup or VTL part of the architecture? Does this storage count in the denominator of your TB/person ratio?

Well there is a lot to consider, as you can see from the outline above, to calculate and consider best-in-class for storage labor ratios. I do not recommend that comparisons to national or vertical markets should be or can be made. These comparisons are never apples-to-apples. There are too many considerations and variations in work tasks, architectures and support function. That is why do you do not see analysts like IDC or Gartner talk about these ratios or publish best-in-class ranges (at least I have not seen these types of reports in the last 4-5 years)

What can be done effectively though is to use TB/person for internal comparisons. From the above list, decide what you want to use in your own Labor Ratio Benchmarks, and then compare progress over time with your own calculation or dashboard. Using the TB/Person approach is reasonable for internal measurement of IT effectiveness. Your comparisons and the elements chosen to record or track will be indicative of what operational support categories are important to you.

There are many things that can be done, to improve the management ratio; I will explore that topic in more depth next time

Trackback URI | Comments RSS