DIFFICULT DECISIONS: In tough economic times, pressure on budgets across all areas of enterprises is demanding hard choices in IT investment. - ConocoPhilips
IN TOUGH economic times, pressure on budgets across all areas of enterprises is demanding hard choices in IT investment, made even more problematic by the exciting possibilities - and potential problems - of new storage architectures.
In spite of the economic climate, there are four key, timeless strategic principles that will hold true through any macro crisis or technological leap forward, allowing enterprises to identify and measure costs, reduce capital expenditure, rein in operating expenses, and grow sustainably.
Virtualisation is the catchword for the new era of outsourced IT services. New storage architectures are coming online at an increasingly rapid rate. Cloud computing, virtual machine (VM) sprawl, capacity-on-demand architectures and other architectures demand a review of existing storage infrastructures, prices, costs and operational methods.
Cloud storage, with its promise of lower costs, and businesses freed to focus on their own core capabilities and priorities, is one of the prime contenders. However, enterprises still need to plan and account for their entire IT architecture costs, including virtualised and cloud-based components, in order to achieve true efficiencies and real ROI (return on investment).
Treating IT services or cloud as just another utility alongside water and electricity has to take account of the fact that a company's IT infrastructure is not a fixed cost for a fixed supply, like the gas or electricity mains: It is a dynamic, constantly changing expression of the company's business processes and information flow.
Four principles of storage economics
Storage economics allows us a financial and economic view of storage decisions and architectures. A very common question, and a very real problem, that IT people hear is "we have to reduce costs." CFOs say "reduce costs" while CIOs care about the technologies, that is, they want more "blinking lights." Most of the time, the solution to these conflicting demands is to buy cheaper disks, servers and applications. Therein lies the problem - they confuse price and cost.
Purchasing low-cost storage solutions does NOT equate to lowering operating expenses or reducing the total cost of storage ownership (TCO).
For companies considering cloud computing and virtualisation, the benefits to the business can be huge, but IT management needs to realise just what it can - and cannot - virtualise, and heads of companies need to abandon the mentality that identifies cost just with CAPEX, otherwise they will come to a cloud-based strategy with wrong expectations about what it can achieve for the bottom line.
They also need to be wary, if they are already using third-party service or infrastructure providers, of whose costs are being saved - their own or the provider's.
IT management should first identify the costs of the existing storage architecture, measure the costs, decide what levers would impact these costs, and then pinpoint who exactly will benefit from the cost savings.
Just transferring costs instead of actually reducing them obviously brings little benefit, and this is one reason why you need to look at all the cost savings and other benefits that could accrue, before taking a leap into the clouds. You need to think like an economist, talk like an accountant, and act like a technologist when driving these changes.
The following four principles provide a timeless framework which can help you make strategic and tactical investments to gain business benefits and successfully control the costs of your storage infrastructure in the long run.
Applying these four principles when any new architecture is considered will help separate the price hype and the long-term operational costs associated with every new architecture.
1) Cost of ownership includes more than price
Remember that price does not equal cost. CAPEX on IT infrastructure has steadily diminished as a share of the total cost of ownership (TCO) of a data storage system over the past few years, from 50-60% to about 20% today.
The other 80% is mostly operating expenditure (OPEX), including labour, maintenance, and power. In fact, unit cost of disc space can be seen as approaching zero, but recurrent running costs of a system are becoming the largest and most critical item of IT TCO. These are the costs that managers need to identify and rein in.
2) There are 34 types of cost
On the principle that "you can't improve what you can't measure," management needs to identify and quantify the actual costs that make up its TCO. Collectively, 34 types of costs have been identified (see attached table) that make up storage TCO some hard or direct, some soft or indirect, some CAPEX, others OPEX.
Not all apply equally to every business, and a cloud-based strategy may reduce only some - it could increase others.
For instance, an imperfectly-executed cloud-based strategy could increase litigation risk costs, or security and encryption costs, while even leaving storage management labor unaffected.
Out of the 34 costs below, hardware depreciation, maintenance and warranty, storage management, power consumption, monitoring, and datacentre floor space - should be directly reduced by a cloud-based solution; software purchase/depreciation and software maintenance could also be potentially reduced.
One should note however, all the other costs could be increased under a cloud-based strategy - if not implemented correctly.
Equally, some of the most important cost savings and advantages can be missed through old-fashioned reflection and thought. The cost of growth is one item where a good cloud-based strategy can truly reap rewards, enabling businesses to scale quickly and expand their IT establishment at a fraction of the traditional outlay.
Businesses too wedded to planning on the basis of their existing needs and infrastructure may overlook benefits for the future.
IT planners need to take a full 360-degree look at all these 34 costs, decide which are most critical for their business processes - present and future - and use cloud-based and in-house resources accordingly.
Costs need to be categorised by costs kept internal after the cloud transformation, and the costs that get transferred to the cloud provider.
3) "Cheaper to Own" - Economically superior storage architecture creates value
The best-value storage architecture is the one that saves most, not the one that costs least. Moving costs off the internal or CAPEX balance sheet may simply load them onto other areas, if a cloud-based system is not implemented correctly.
Storage TCO has to be calculated for the long term, as high operating costs will soon start to indicate. This architecture may not be the cheapest to buy, but it should be cheaper to own, and IT planners need to carefully consider how cloud components will actually affect TCO. Some of the key ingredients include:
Virtualisation of volumes, file systems, storage systems;
Dynamically tiered storage;
Intermix storage;
Thin provisioning;
Power down disk, MAID;
Multiprotocol SAN storage;
De-duplication, data compression;
Integrated archive; and
Management, policy-based provisioning
Storage virtualisation, together with tiered storage and dynamic (or thin) provisioning - has been repeatedly demonstrated to reduce TCO by 20-35% over older in-house or tiered-island storage architectures.
Out of the 34 types of cost, the ones which are most affected are waste, migration, copies and labour.
4) Econometrics will show you the way
Econometrics - an economic measuring system to quantify costs and track progress in reducing them - has to be put in place alongside good storage architecture, to map storage initiatives or investments to areas of measurable costs, and use this information to design, prioritise, and roadmap activities, based on their projected potential to reduce costs and support business needs.
Cloud-based systems in particular need econometrics to keep track of costs that are now often out-of-house and not readily identifiable by traditional internal accounting.
Obviously, there is a risk of getting lost in the cloud - losing not only data, but also money. A short-sighted focus on slashing CAPEX may only increase this risk, as technological change is clearly pushing IT towards longer-term time horizons in calculating TCO.
As IT departments find themselves caught between the two imperatives of technological advances and cost pressures, managers need to decide the basis for handling and balancing these imperatives.
The Four Principles of Storage Economics at least give a sound foundation for a storage architecture that really can rise above the clouds.
(Johnson Khoo is managing director of Hitachi Data Systems Malaysia)
The 34 types of cost
1. Hardware depreciation and leases
2. Software purchase or depreciation
3. Hardware maintenance or warranty
4. Software maintenance or warranty
5. Storage management - e.g. labour costs for provisioning, tuning, load balancing, troubleshooting and upgrades
6. Backup and disaster recovery - e.g. labour costs for backups and restores, disaster recovery planning and testing
7. Data migration, re-mastering labour costs
8. Data mobility - time and effort to move data to different tiers or archive solutions during the data lifecycle
9. Power consumption and cooling costs
10. Monitoring - SNMP, NOC and operations consoles for the storage, SAN and backup infrastructures
11. Datacentre floor space
12. Provisioning time - business impact of the time needed from when the request is made until capacity is presented to the host
13. Cost of waste (Two types: Usable and not allocated, and allocated and not used)
14. Cost of copies - database management systems (DBMS) and other applications often require multiple copies to be made
15. Cost of duplicate data - overhead associated with several copies of the same data
16. Cost of growth - every storage architecture has a cost of growth. In high-growth environments with the wrong architecture, the cost of growth can be acute
17. Cost of scheduled outage (microcode changes, capacity upgrades)
18. Cost of unscheduled outage (machine related)
19. Cost of unscheduled outage (people and process related)
20. Cost of disaster risk, business resumption
21. Recovery time objective and recovery point objective (RTO and RPO) costs - business impact costs resulting from the time it takes to return to a recovery time (or point) after a system failure or backup recovery
22. Data loss
23. Litigation, discovery risk - legal risk and e-discovery time costs associated with lawsuits
24. Reduction of hazardous waste - primarily an EU cost due to regulations such as RoHS. Noncompliant hardware may incur an additional tariff for disposal of the asset
25. Cost of performance - impact to the business (good or bad)
26. Backup infrastructure - includes backup servers, media servers, tape libraries, drives, etc
27. Backup media - local and remote media costs for backup; recurring and capacity related costs
28. Cost of risk with backup windows - business impact of shortened or limited backup windows
29. CIFS- and NFS-related infrastructure - filers, gateways and the necessary software to provide file servers and shared services in the enterprise
30. Local and remote data circuits - dark fibre used for SAN extensions, remote replication and the associated software
31. Storage area networking - dedicated Fibre Channel, iSCSI or NAS connection infrastructures. This includes routers, gateways, host bus adaptor switches and directors
32. Noncompliance risk (archive, data retention) - several legal and legislative requirements (HIPAA, Basel II, Sarbanes-Oxley, carbon emissions), non-compliance with which can incur fines, negative publicity and criminal prosecution
33. Security and encryption costs for protecting data and storage infrastructure
34. Procurement - costs associated with time and effort required to acquire hardware and software, including preparation, review, negotiation, selection and certification