europe SNIA
www.snia-europe.org
This model is heavily challenged by smart disk arrays with ‘virtualization inside’ as vertically integrated models proposing simplicity while addressing virtualization business benefits.
when de-commissioning whole disk arrays. Horizontal virtualization supports projects measured in days even for swaps involving heterogeneous disk controllers.
‘Reduction of burden associated with storage moves, adds and changes’ is pretty much comparable in both architectures. The adoption and the usage of variable size LUN brought by virtualization does simplify storage administration burden regardless of the vertical or horizontal virtualization model. The difference might reside in the initial data-placement effort for which organizations used to spend sizable time measured in weeks to assess which array and tier should be used for a new application. Today, data mobility within arrays is widely implemented and this reduces the time spent on tiering decisions; only array selection tends to remain an issue. Virtualization in the infrastructure goes beyond this as data mobility across arrays allows releasing applications in nursery pools for observation before moving them to their appropriate place and revisiting this later.
For the sake of completeness, we also need to mention hybrid models where smart disk arrays may also federate external disk arrays providing their own disks for capacity usage by storage pools.
These 2 major models are addressing the end user wish list with various capabilities depending on their architecture and implementation. Let’s look at each of them and discuss how they address user requirements.
‘Reduction of time and complexity of moving data from one system to another’ is clearly best addressed by ‘appliance based’ solutions. There architectures are designed for this even if some solutions may require a short service interruption when switching data from back end disk array to other ones. Disk array solutions are mitigating this disadvantage by providing scenarios based on replication, ‘hyperswap’ techniques and/or host based mirroring techniques. Key differences lay in the amount of disruption to copy services, interruption of service and most importantly in the duration of projects
16
www.snseurope.info I October/November 2011
‘Implementing tier storage strategy based on heterogeneous storage systems’ is definitely better suited for horizontal virtualization at the expense of implementing an extra hardware and software layer. Vertically virtualized disk arrays are proposing ‘tiering in the box’ with data placement automation engines with various granularity (LUN or sub LUN) and perimeters (in-pool or between pools). Vendor algorithms are still emerging with limited field experience and their relevance for client applications may vary over time. One should not take for granted that automated means optimum especially because algorithms are evolving based on actual client’s usage. Similar techniques are also available for selected ‘appliance based’ solutions. A key aspect for tiering solutions is not only the optimization of data placement but also the ability to support decisions of a future tier purchase based on actual infrastructure load. Tier storage is clearly a CFO desire and the best way to implement this is to start by optimizing future planning and spending. One should also consider that virtualization appliances come with hardware rich in cache adding IO processing capacity to discrete disk arrays and, this represents either sizable performance increase benefit or capex reduction savings versus a vertical alternative.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44