Sometimes during the sales process the wording which accompanies the sale of a high availability (HA) product can change between providing disaster recovery (DR) and HA. Are they both the same? No, says Chris Hird of Shield Advanced Solutions (Canada). ‘HA is more about removing planned downtime, while DR is all about recovering after an unplanned event.
‘HA, because of its nature, has the ability to provide DR as well, especially when the system has been segregated geographically,’ says Hird. ‘But having the system separated in this way also brings extra challenges if you want to manage the planned outages; switching users across network links has to be well thought out and the switch process properly designed.’
It’s been said that 95% of system outages related to System i are due to planned events, but the extent of the disruption caused by an unplanned outage can be more devastating to the business. ‘Unless you’re after the ability to switch users regularly, HA may be overkill and DR or a derivative could be better,’ says Hird.
‘I’ve been involved in many HA sales,’ Hird adds. ‘Most have been made because the company wants to protect its data and be able to switch quickly and regularly. Yet many of those customers fail to switch at all until a disaster strikes. They then find the environment has changed so much the recovery is more painful than it should be. Many others have not bothered to go for a HA solution because they want data protection, not HA. Spending thousands of dollars in an attempt to provide that data protection has put them off moving forward.’
Konrad Litwin, managing director at eveLogic, says: ‘Something that should be considered in relation to any DR plan is secure backups with encryption. Many organisations have secure systems, but as soon as their tapes go off-site or are sometimes easily accessed on-site, they are open to being read by anybody looking to get a heads-up on the data. Select data should be encrypted on the tapes,’ says Litwin. ‘You then have the requirement to keep your encryption key as part of the DR plan, so once you get to your back-up site you can restore the data.
‘Robot/SAVE encrypts data as it is saved to tape, making back-up tapes as secure as the system,’ continues Litwin. ‘Users specify what to encrypt – libraries, individual objects, object lists, IFS objects or Domino databases and the encryption process becomes part of the overall back-up strategy. Restoration of encrypted data is transparent. If a tape is lost or stolen, the encrypted files are unreadable. Encryption helps compliance with government regulations/privacy laws, for example SOX, HIPAA and PCI.’
Business growth depends on IT uptime. Recent research by the Aberdeen Group showed that Best in Class (BIC) companies – ie companies with the overall ability to recover critical applications within a short window and year-on-year improvement in the ability to recover data – shared a number of common characteristics. The findings showed 53% of BIC companies had HA strategies in place for over three years and 63% used failover software.
Combined effort
However, a BIC implementation of a HA strategy requires more than just technology; it requires a combination of actions, organisational capabilities, and enabling technology. Enter System i and solutions providers. Nigel Adams, UKISA System i product and sales programme manager, says: ‘In the age of the internet and 24x7 business, availability is an issue of ever-increasing importance. Regulatory issues have also greatly increased the focus that businesses have to have on this.
‘System i and i5/OS have always had a well-justified reputation of being a system offering high levels of availability. But over the years IBM has added many new hardware and software capabilities that enhance the availability of the system,’ says Adams. ‘More recent additions include Switchable Independent Auxiliary Storage Pools allowing storage to be switched from one system to another, and Cross-site Mirroring providing mirrored copies of independent storage pools to a remote location. These are powerful capabilities which can be used to provide high levels of availability.
‘We’ve also just announced a product preview of V6R1, the next release of i5/OS, and there will be functionality in this new release addressing HA and DR,’ adds Adams. ‘The product preview states, among other things, continued investments in i5/OS business resiliency, including enhancements to cross-site mirroring and further simplification of system object replication through the administrative system domain. Over the years, we’ve relied on HA Business Partners to provide logical replication, and the ability to rapidly recover applications and data from a system failure. This continues to be the case, even though the functionality offered within i5/OS in this area does increase with every release. The HA Business Partners provide a capability that builds on i5/OS and offers real HA in terms of recovery time and recovery point to System i customers.’
Vision Solutions’ executive vice-president and chief technology officer Alan Arnold says: ‘We have high expectations for systems availability and there are numerous reasons why companies should consider protecting their business. If systems are down, the majority of employees cannot get their jobs completed. At the end of the day, the reason is irrelevant. Most companies only have seconds to complete transactions before they risk losing customers. All companies’ reputations are negatively affected if their systems are not available. For big companies a system failure means front-page news; for small shops, customers won’t forget. Savvy organisations understand how to use technology to their competitive advantage and a key component of any technology architecture includes HA.
‘Another important reason for having HA is regulatory issues,’ says Arnold. ‘In most parts of the world, public companies have regulatory responsibilities to protect and manage their data assets. Everything from financial records to patient files must be protected. Management of public companies can be held financially responsible and no excuses are acceptable for missing or wrong information. Thus, management teams must understand their fiduciary responsibilities for their specific industry and have an appropriate data availability strategy in place.
‘Finally, companies of all sizes must prepare for unforeseen disasters. At Vision, our solutions have helped our customers through terrorist attacks, hurricanes, earthquakes, and floods,’ says Arnold. ‘With the UK experiencing the worst flooding in over 60 years, many companies will experience serious issues – especially if they have not protected their IT systems. Some may never recover, which is unfortunate and could have been avoided with the proper HA solution. At Vision Solutions, we help companies be prepared so that they can switch their computing environment to another facility, city, region, or even another part of the world should the need arise. Every company should have HA. Not being prepared can cost millions; in fact, the ultimate cost is going out of business altogether.’
Bouncing back
Jonathan Woods, a strategic architect at Logicalis, has an interesting take on HA: ‘In an “always-on” global economy, a company’s ability to bounce back from a setback and the speed at which it is able to re-establish “business as usual” is now critical to its survival. It’s generally acknowledged a company’s most valuable assets are its people and its knowledge, and that the two are interdependent. This is particularly true in a fast-changing risk/disaster environment, where humans are unequalled in their ability to share and use knowledge to deal with new or unique problems as they arise.
‘The modern approach to risk management emphasises the need to provide staff with access to data rather than a desk. Such an approach requires businesses to rethink the traditional methods of continuity planning that focus predominantly around data and office location backup. The priority should now be to protect and enable staff access to data, as well as focusing on protecting centralised infrastructure, therefore making HA key to business continuity,’ he says.
Logicalis calls this new approach Continuity by Design. ‘With this strategy in place, businesses are enabled to work wherever they are located and therefore equipped to provide ‘business as usual’, adds Woods.
Quattro Consulting’s managing director Glenn Robinson says: ‘HA and DR have always been high on the agenda for i5/OS customers and the recent acquisitions in the marketplace will ensure customers have a wide selection of HA/DR package options from a small number of suppliers. For a number of releases now, IBM has been putting more and more HA functionality into i5/OS, such as Cross-site Mirroring, Switchable Independent ASPs and Clustering. These are becoming more sophisticated and easier to deploy with each release.
‘I believe that from V6R1, we’ll start to see more customers utilising the i5/OS functions for HA and DR, but there’ll be some customers who still require the additional functionality and application integration provided by the ISP packages,’ he says.
‘The clustering technologies provided by i5/OS have taken a long time to gain acceptance in the System i customer base because of their complexity and lack of functionality,’ he adds. ‘This is now starting to mature and will gain much wider acceptance by customers. This is similar to the take-up of LPARs on AS/400 and System i; this was slow at first and seen as “just for the big boys”. Now we see the majority of customers using partition technology and i5/OS clustering technologies will start to move in the same direction.’
Within the HA and DR pool, users may be forgiven for being confused or perplexed over the myriad of mergers and collaborations going on in the vendors’ camp. The battle for the System i HA market took yet another unexpected twist this summer when IBM announced it intended to purchase DataMirror for around $161m. It’s expected discussions about this will be made this month once all issues have been agreed.
DataMirror was one of the original ‘big three’ System i HA vendors, along with Lakeview and Vision Solutions. Lakeview and Vision have now been merged along with second-tier HA provider iTera, under the aegis of American private equity firm Thoma Cressey Bravo, leaving New Zealand’s Maximum Availability and French firm Traders as the only independent companies in this particular market sector.
Meanwhile, Paul Ranson, DataMirror, sales, says: ‘The key thing a company needs to do is assess what are the real business recovery needs required by them to comply with regulations; for example, from the likes of the FSA for financial services, or the security and well-being of their business for them, the owners or shareholders and, most importantly, the service to their customers. How many times have you been on the phone/internet to order some products or services to be told “the computers are down at present” and “can you call back later?” – do you? Or do you just go elsewhere?’
‘While Maximum Availability can deploy a rapid implementation in a matter of hours, it also recognises that the reality of HA software is that it's not a one-size-fits-all solution, with each implementation needing to be tailored to the particular customer’s need,’ says Andrew Mansfield, vice-president of marketing and business development at Maximum Availability. ‘Just the same, the company still emphasises the ability to quickly and cost-effectively evaluate (initial sales phase, including demos, trials, and pilots) in days rather than weeks or months. The full implementation shouldn't take a whole lot longer than that. The days of multi-month implementations are over.
‘Maximum Availability has also developed new functionality – over 50% more in the last release – into its *noMAX product line to compete in an increasingly tough market and further significant development continues. A second major new release of *noMAX is expected by the end of the year. Key product elements the company is working on include boosting the product's GUI, adding more autonomics, and heightening integration with SAP environments.
‘Then there's the acceptance of remote journaling as a replication mechanism and the relative affordability of hardware the System i Express Models 515 and 525 have brought, not to mention the Capacity Backup (CBU) On-Demand offering from IBM, which drops the price of an HA solution even more,’
adds Mansfield.
Record sales
Maximum Availability has invested considerably recently bolstering capability in all key regions by placing more *noMAX sales and technical support on the ground, signing up new business partners and making inroads with IBM across the globe to drive targeted market development activities. Global sales in the last quarter were reported at record levels and with its continued efforts combined with more opportunity being generated from recent convergence in the System i market, the company is bullish about the future.
Hitherto there have been no alternatives. If you need to protect your data and be able to recover to within the last few transactions, you had no option but to go for a HA solution, Chris Hird points out. ‘That’s now changing with more BPs offering hosted services where the cost of the HA product is being split over many environments. New products are coming to market which use remote journaling technology to store that data on another system and apply frequently. If you just need data protection without the need to switch regularly, the initial cost and ongoing management costs may not stack up,’ says Hird. ‘I’m sure IBM will have more hardware-based solutions in the future which will allow smaller customers to gain benefit from it.’
Frank Booty, industry reporter
