Drobo Technology Innovation

Transcription

1 Drobo Technology Innovation Drobo 2460 North First Street, San Jose, CA DROBO Copyright 2011 Drobo, Inc. Data Robotics, Drobo, DroboElite, DroboPro, Drobo, BeyondRAID, and Smart Volumes are trademarks of Data Robotics, Inc., which may be registered in some jurisdictions. All other trademarks used are owned by their respective owners. All rights reserved. Specifications subject to change without notice. EB July 2011 Drobo Technology Innovation Page 1

2 Contents Introduction... 3 BeyondRAID... 3 Challenges of Traditional RAID... 4 Thin Provisioning and Reclamation... 6 Mixed Drive Utilization... 7 Automatic Protection Levels... 8 Virtual Hot Spare... 9 What is Data Aware? Drive Re-ordering How BeyondRAID Works Drive Handling Drobo Disk Pack BeyondRAID Zones BeyondRAID Update and Rebuild BeyondRAID Clusters Data Aware Background Data Checking Drive Error Handling Automated Data-Aware Tiering Being Data Aware One Pool, Multiple Tiers Fully Automated Tuning Tiering Without SSDs Just Add SSDs Really! Industrial Design and User Interface Storage Without Screws As Easy as a Traffic Light Storage Designed for You Advanced Management That s Not Advanced Automatic iscsi Configuration Benefit and Cost Savings from Drobo Technology Fewer Drives Fewer Arrays Lower CapEx Lower OpEx Fewer SSDs Summing Up Appendix A: BeyondRAID vs. Traditional RAID Appendix B: References Drobo Technology Innovation Page 2

3 Introduction Storage has traditionally been difficult to understand and manage, especially for non-storage experts. Tasks such as pooling drives, scaling capacity, and protecting from data loss are complex tasks for individuals and businesses that require high-capacity and reliable storage. Also, the explosion of rich media data requires more sophisticated digital storage. Legacy solutions can be too complex or too limited for current needs. Drobo s unique technology in all our products delivers breakthrough ease-of-use, affordability, and value so that individual professionals and businesses will have the best storage experience ever. Our patented technologies are the basis for enabling a unique combination of simplicity and storage sophistication. Our products offer a stark contrast to legacy storage, with distinct advantages in Ease-of-use, Affordability, data Safety, and Expandability (EASE), with the overall design that addresses the needs of workloads in the areas of media, virtualization, and cloudattached storage. Three unique Drobo technologies provide users with a truly modern storage experience: 1. BeyondRAID breaks down the barriers of traditional RAID to greatly simplify storage deployment and management and protect data and files 2. Automated Data-Aware Tiering is enabled by BeyondRAID and allows storage to optimize storage automatically based on the type (or tier ) of data being stored. 3. Industrial Design and User Interface make pooling the drives in a single chassis easier then could ever be imagined. BeyondRAID RAID (Redundant Array of Independent Disks) technology was first introduced as a concept in the late 1980s. It was designed to provide larger logical storage containers that were capable of protecting data from a random disk drive failure. Since then, a majority of businesses and many individual professionals have leveraged RAID-enabled products to store large amounts of valuable data. Figure 1. RAID allows disk drives to be pooled into logical containers (arrays). Drobo Technology Innovation Page 3

4 Drobo was founded to create storage that breaks down many of the inherent barriers of deploying traditional RAID. The result was a new and innovative technology called BeyondRAID that delivers all of the benefits of traditional RAID providing greater capabilities and drastically simplifying implementation and management and as a result, leaving the limitations behind. This nextgeneration of storage virtualization technology brings together enhanced protection, reliability, expandability, and ease-of-use. Challenges of Traditional RAID To deploy traditional RAID-enabled storage, you first had to learn about RAID and understand the necessary prerequisites before the storage can be consumed by a computer and its applications. This raises the question: should all individual professionals and IT administrators have to learn what an enterprise storage administrator knows to deploy RAID-level storage? And to the observation that there must be an easier way to deploy storage! Figure 2. There must be an easier way to deploy storage? First, you need to decide up front how many disks should be in a RAID group. Do you need 2, 3, 4, or even more disks? What capacity disks should you purchase? This is a critical decision because with most arrays, you can t grow the RAID group with additional disks after it s already in use. It s challenging to predict how much storage space you ll need two years, one year, or even six months from now. So if you plan to deploy traditional RAID, it s smart to invest in a lot of storage initially so you can grow into it over time. You just have to accept the added capital cost. When the RAID group is full, you must create a new RAID group to store more data. This is because many arrays do not offer functionality to grow a RAID group. And if you re out of empty drive bays, then this new RAID group needs to be on a different storage array. If you don t want to manage your data on different RAID groups, you ll need to set up a new array that has more capacity, migrate the data from the old array to the new array, and then figure out how to repurpose the old array. It s easy to see how this could be time consuming and costly. Drobo Technology Innovation Page 4

5 Often, you must determine the RAID protection level when you deploy, because many arrays don t allow you to change RAID levels. Even when they do, the entire RAID set must be rebuilt. This can sometimes take days and even weeks depending on the size and number of drives in the RAID group. When changing RAID levels, the volumes, or LUNs, may not even be available for use. Can you handle that downtime? When a drive fails in a traditional RAID array, the RAID group is in a degraded state in which the data is not protected. Some arrays even prevent writes when they re in a degraded state. The array will be this way until the failed drive is replaced with a new one. Hot spare functionality was created to automate this, but what are hot spares but unused storage that just consumes a drive bay and waits for a failure? If you are not familiar with RAID arrays, you may be thinking, Wait a minute you re telling me that I have to burn multiple drives for RAID protection and then I need a hot spare to sit there in case a primary drive fails? How much usable storage will I actually have? That s a good question. Figure 3. In this 12-drive array, each of three RAID groups has their own parity drive plus a hot spare in case of a drive failure. 4 of the 12 drives are not used to store data. On top of all of that, in most RAID arrays, you cannot use different sized drives in a RAID set. This means that if you want to leverage larger drives, you either need to create a new RAID group containing that larger drive or you need to migrate your data from the old RAID group to a new RAID group and then repurpose the old drives. Once you get past the challenges of RAID groups, you have to carve up the available storage by creating volumes, or LUNs, and assign those volumes to a specific server. It s like renting storage space in a self-storage facility space is assigned to you and whether or not you use it, it cannot be assigned to anyone else. A single computer or server owns the entire capacity of that volume. That s a lot of free space tied up by individual computers yet another challenge that makes storage provisioning difficult. Drobo Technology Innovation Page 5

6 Figure 4. Rather than pulling from a common pool of free space, RAID groups are carved up into volumes (LUNs) and assigned to servers. Now let s see how these challenges are addressed with innovative Drobo BeyondRAID technology. Drobo BeyondRAID technology overcomes all the challenges of traditional RAID described above by fundamentally changing the way block data is stored. Thin Provisioning and Reclamation One of the toughest challenges of storage management is to decide how much storage space is needed for a computer or application. If you allocate too little you may run out of space, too much and you waste valuable storage resources by tying it up to one computer. Thin Provisioning is a fundamental capability of BeyondRAID, allowing you to allocate the right amount of storage for today, while allowing you to expand capacity by just adding a drive to the Drobo or replacing an existing drive with a larger one. With Drobo, this expansion is nearly instant and occurs when a new or larger drive is inserted without any disruption in service. If you are running low on space, all you need is another hard drive to add to the Drobo. Figure 5. Adding additional drives instantly increases usable storage capacity, in this case, from 1.5 TB to 2.5 TB (assuming single disk redundancy). Drobo Technology Innovation Page 6

7 While 4- and 5-bay Drobos often contain one thinly provisioned 16 TB volume, 8- and 12-bay models allow for user-controlled volume creation. These volumes (or LUNs) are called Smart Volumes and they pull needed storage from a common pool (thin provisioning) and then return deleted blocks back to the common pool (thin reclamation), preventing free space from being tied up by any particular server. Space allocation is automatically managed, which helps maximize storage resources. You can try this out for yourself. If you have a large file on a volume and then you delete it, you will see the free space of the Drobo grow, making that freed up space available to any other volume. Even enterprise-class arrays with thin provisioning do not return consumed storage back to a common pool at this granular of a level. Figure 6. As volumes consume storage, they pull from a common pool of free space and when storage is freed up, it is relinquished back to the same pool. Mixed Drive Utilization With Drobo you can insert any drive of any size and it will be used. With traditional RAID (RAID 5, for example), all drives must be the same size to be utilized. What if you have a set of three drives in a RAID group and you want to expand? With traditional RAID, you must migrate data off your drives, add a new drive that is the same size as the others, reformat the RAID group, and then migrate the data back onto the drives. Not only does this take a lot of management time, it incurs potential downtime; you need available storage space in another array to temporarily store the data. With Drobo, if you re running low on space and you have an empty drive bay, just add another drive. Even when your Drobo is full, just remove the smallest drive and replace it with a larger one. Figure 7. If mixed drive sizes are used in traditional storage, drives of the same size must be in their own RAID group, drastically reducing storage utilization and flexibility. Drobo utilizes mixed drive sizes in a single group. To best learn how mixed drive utilization works in a Drobo, play with our interactive Capacity Calculators. They allow you to simulate putting drives of mixed sizes in a Drobo to see exactly how the drives will be used. Drobo Technology Innovation Page 7

8 Automatic Protection Levels A Drobo always sets itself up to protect from random drive failure. Unlike traditional arrays, there are no options in a Drobo to turn off drive protection (nor would you want to). In a Drobo, the only options are Single Disk Redundancy and Dual Disk Redundancy. You can switch between these protection levels on the fly and Drobo adjusts the data on the drives appropriately. About RAID Levels With a Drobo, you do not need to learn about all the various RAID levels for drive protection/performance/capacity trade offs, but it doesn t hurt to know a little: ü RAID 0 drives are striped together and provides no data redundancy to protect from a drive failure. While performance often increases with RAID 0, reliability decreases each time a drive is added to the set. ü RAID 1 drives are mirrored together and provide data redundancy between two or more drives. Most often two drives exist in a mirror and performance is similar to that of a single drive. ü RAID 5 drives are striped together like with RAID 0, except parity data is created and spread across all drives in the set to prevent from a single drive failure. This RAID level is common because it scales while keeping the cost of drive redundancy down. ü RAID 6 drives are striped, with parity data spread across all drives similar to RAID 5, but in this case, the parity data is stored twice. This protects against two drive failures. Often, RAID 6 has decreased write performance compared to RAID 5 because of the added time to calculate and store an additional set of parity blocks. There are other RAID levels such as RAID 4, RAID 0+1, and RAID 50 all of which makes determining the proper RAID level even more complex. Drobo Technology Innovation Page 8

9 BeyondRAID only has two protection levels: ü Single Disk Redundancy provides protection from a single drive failure at a given time (similar to RAID 5). Storage capacity is maximized with this option. ü Dual Disk Redundancy provides protection from two simultaneous drive failures (similar to RAID 6). You invest by consuming an additional drive but gain the added protection. Even when the Drobo is automatically rebuilding from a previous drive failure, you are protected from another drive failure, ensuring that your data remains available and accessible. Drobo guarantees that the best algorithm is used to protect from drive failure depending on the number of drives that are inserted without any user configuration or intervention. There is no need to learn about RAID levels or how mirroring or striping works. For example, if you enable Dual Disk Redundancy and have only two drives inserted, Drobo will operate, but will tell you to insert another drive to operate in a fully protected state. Virtual Hot Spare A Drobo utilizes all drives all the time with no drives sitting idle waiting for one to fail. With some traditional RAID systems, a drive can be marked as a "hot spare," waiting on standby for a drive failure so it can take over. That's just an unused drive that you have to pay for taking up a valuable drive bay. Figure 8. With a Drobo, no wasted hot spare is needed for automatic rebuild in the event of a drive failure. If a drive fails in a Drobo and there is enough free space available, Drobo will automatically move data around to return to a protected state. No user interaction is required. This is called a virtual hot spare because the unused space in a Drobo IS the hot spare. Once the Drobo automatically recovers from a drive failure by using the free space available, if sufficient amount of free space is still available, the virtual hot spare will still be there. With traditional storage, the admin would have to replace the failed drives and elect a new hot spare. Drobo Technology Innovation Page 9

10 What is Data Aware? Drobos are data aware, that is, they intelligently handle data for added resiliency, performance, and management. This is different than just storing data at logical block addresses like RAID. Being data aware informs the capacity gauge on the front of the Drobo, quickly recovering in the event of a drive failure, enabling intelligent tiering of data, and much more. Figure 9. Data is analyzed in-flight as it enters a Drobo and is intelligently placed on the drives. As data is written to the Drobo, it is analyzed in flight and intelligently placed on the media to maximize protection and performance. Drobos are also aware of the data placement inside the file system. So Drobo can display an accurate capacity gauge without even being connected to a computer or to enable Smart Volumes as described earlier in this chapter. Drive Re-ordering With traditional RAID groups, drives have to be in a specific order. With disk packs in a Drobo, ordering does not matter. This allows Drobo disk packs to be moved to another Drobo and even reordered, while still functioning. Try that with RAID and see what happens to your data! Figure 10. Unlike traditional storage, drives in a Drobo do not have any specific number or order, so they can be placed in any slots and function normally. It s one less thing to worry about. Drobo Technology Innovation Page 10

11 How BeyondRAID Works Now that we ve looked at how BeyondRAID compares with traditional RAID and the value that it brings, let s see how BeyondRAID works. Figure 11. The sections of the RAID stack are drastically reinvented in a Drobo and it all has to do with how drives are pooled (disk pack) and the way data is stored (BeyondRAID virtualization). Drive Handling As mentioned in the previous section, drives are not tied to a specific bay or order. Using drivespecific information, a unique identifier is assigned to each drive. Because drives in a Drobo are identified by their unique identifier, they can be placed in any slot position, and reordered, while still being recognized as the same logical drive. Figure 12. Drives in a Drobo are logical in nature because they have a unique identifier, allowing them to be placed in any bay and function normally. Drobo Technology Innovation Page 11

12 Drobo Disk Pack A disk pack refers to the group of drives in a single Drobo chassis. Unlike RAID groups or disk pools in a traditional array, each Drobo chassis has only one storage pool, which is the disk pack, This makes managing drives in a Drobo very easy. The disk pack can be loaded into a Drobo with the drives in any order and no matter what, behave as one storage pool. A disk pack missing drives is still accessible provided there is still a copy of all the data. A disk pack missing a single drive will have lost some or all of its redundancy, so it will be in a degraded state. When a disk pack is degraded, any drive that would cause a loss of the remaining copy of some data on the pack is a critical drive and cannot be compromised or removed. In general, loss of a critical drive will cause the system to reboot and wait for the critical drive to be replaced. Since a Drobo is always set up to protect data, this would occur only if two or more drives are missing from the disk pack. Figure 13. A disk pack missing a drive still functions but is in a degraded state and causes a rebuild to occur if enough free space is available. If a drive is inserted into a booted Drobo, it will automatically be absorbed into the disk pack. If data exists on a drive, it should be copied off or backed up before inserting that drive into the Drobo because it will be automatically reformatted. This functionality allows drives to be inserted into a Drobo to scale available storage without any additional administration. Drobo Technology Innovation Page 12

13 BeyondRAID Zones Zones are the core unit of redundancy in a Drobo disk pack. Regions from multiple drives are combined to create a redundant zone. Figure 14. Regions from multiple drives are combined to create a redundant zone. Zones are either single disk redundant (SDR) or dual disk redundant (DDR), meaning that there is either one or two additional copies of the data, allowing the loss of either one or two copies of the data before a fatal loss of data occurs. Turning on and off dual disk redundancy is the only data protection option you need to worry about in a Drobo which makes it very easy for you. Different types of zones using different levels of protection (similar to RAID levels) are automatically created within a disk pack to use all available drive space in the best possible way. This is what allows a Drobo to use drives of different sizes. Figure 15. Zones of different types and sizes can use regions from any drive, optimally allocating available capacity. Zones are always stored redundantly, even if the minimum number of drives to put a disk pack in a fully protected state is not present. For example, if only one drive is present in a Drobo, two copies of the data are kept on the same drive. This provides some protection against corruption on one part of the drive, but the protection is minimal. Drobo Technology Innovation Page 13

14 BeyondRAID Zone Types Different zone types can be used concurrently by Drobo to maximize use of drive space, optimize performance, and allow dynamic adjustments to changes in the topology of a disk pack (from adding or removing a drive or drive failure). These are the different types of zones that BeyondRAID uses automatically: ü Self mirror - two identical copies of the data stored on the same drive ü Mirror - two identical copies of the data stored on different drives ü Mirror 3 - three identical copies of the data stored on different drives ü Stripe - two copies of the data stored striped across multiple drives with the redundant copy stored as parity data (stripe + parity). ü D Stripe - three copies of the data are stored striped across a number of drives, with the dual redundant copies stored as parity data (stripe + dual parity) ü M Stripe - two mirrors striped together for low-overhead writes in single redundancy disk packs. ü M3 Stripe - two Mirror 3s striped together for low-overhead writes in dual redundancy disk packs. BeyondRAID Update and Rebuild When the topology of a disk pack changes, BeyondRAID automatically creates new zones and converts zones between types to incorporate new drives into a disk pack or to return to a fully redundant state. Even when a region of a drive is marked unusable, a rebuild will fix a zone. These operations occur without the servers losing any access to the storage. Each zone is updated or repaired by either a complete re-layout, in which a new, fully redundant zone is created or a patch, in which case, replacement regions are allocated to the zone and the lost regions are reconstituted in place. In addition, if new drives are added to the disk pack, Drobo may rebalance a zone so that it is organized more efficiently across the available set of drives. This involves creating a new zone with new regions, copying the data to the new zone, and then freeing the old regions. A final mechanism for updating and rebuilding is called fast re-sync. If a drive is removed and replaced before the system is rebooted, Drobo keeps track of which zones have been written to it. Drobo Technology Innovation Page 14

15 If the removed drive is re-inserted and the system is still degraded, the zones that haven t been changed from their old location can be repaired, and just the zones that may have changed are relaid out. This can vastly reduce rebuild times. As zones are used on demand, Drobo is aware of how much drive space has been used. In a rebuild in which a drive has been used or replaced, only zones actively in use need to be rebuilt. The number of regions in use by zones will be approximately the same as the amount of data stored on the system, meaning Drobo doesn t need to rebuild the parts of the drive that are unused. BeyondRAID Clusters Looking one layer deeper, the zones are carved up into 4K clusters and are naturally stored redundantly. Zones are allocated automatically to store new clusters. These clusters inside the zones are the global storage pool that is used by all volumes (LUNs) in a Drobo. Figure 16. Clusters consume available space within zones. To enable Automated Data-Aware Tiering (discussed in the next section), two pools of zones are managed: a bulk pool, intended for streaming writes and a transactional pool intended for small, transactional data that needs to be randomly accessed. This allows data to be handled according to its properties instead of treating all data in the same way. Data Aware It is trivial to allocate clusters on demand when data is being written to a Drobo, but storage arrays are not naturally aware of when data is deleted from a file system. In a thin provisioned system such as Drobo, this is important information to prevent leaking clusters, or losing free space to a specific volume over time. Figure 17. Instead of free space being tied up a volume, a Drobo is data aware: free space is pulled from a common pool and returned to that pool when the file system deletes data. Drobo Technology Innovation Page 15

16 Drobo understands and monitors what goes on in the file system to track when logical block address (LBA) ranges are freed up. It can then de-allocate the associated clusters and return them to the global pool of free space. This technology allows the Drobo to be very efficient and free space is not tied up by any specific volume or the assigned server. Background Data Checking Consider backing data and files up to tapes and storing them on the shelf for a decade untouched. What are the chances of a perfect restore when you really need it? Only a small percentage of data stored on an array is regularly accessed. Large amounts of disk surfaces are only occasionally or ever read prior to a rebuild event, which may require access to a redundant copy of the data that is no longer accessible, causing data loss. Background checking, or scrubbing, solves this problem by periodically reading zones on the drive to ensure that the data stored in them is not degraded. If a read operation fails, the zone is repaired via the usual mechanisms and the failing region of the drive taken offline. Many traditional RAID arrays do not perform any background scrubbing. Scrubbing is a background operation and occurs only when the system is not under high load to ensure that scrubbing and similar processes don t impact system performance. Figure 18. When the host server is not accessing its volume on the Drobo, background scrubbing on the zones occurs to ensure data reliability and to proactively repair issues. Drive Error Handling Drobo actively tracks drive errors, and if too many errors are recorded or if the rate of errors is judged too high, a drive can be expunged or rejected. An expunged drive is permanently removed from a disk pack and will not be used again by the Drobo. When this occurs, the slot LED of an expunged drive flashes red to indicate that the drive needs to be removed and replaced. Drobo tracks drive errors, timeout, and start failures, with the goal of expunging a drive before fatal errors, which put the data stored at risk, occur. Drobo Technology Innovation Page 16

17 Automated Data-Aware Tiering Automated Data-Aware Tiering is currently available only for the Drobo B1200i. Over time, additional Drobos will incorporate this technology. How do small and medium business (SMB) IT administrators effectively use solid state drives (SSDs) to accelerate their applications? They can t. Ether the array is too expensive (more than $50,000) or, even if the SMB-focused arrays allow SSDs to be used, they must be in their own pool and separate from the rest of the data. Data must then be manually sent to the fast volume (SSD) or to the normal volume. How is the IT admin supposed to figure out what data is transactional in nature, that is, what data should be sent to the fast volume? And often data has mixed characteristics, making this operation even more confusing and complicated. To get the most out of expensive SSDs, the solution must be intelligent and fully automatic the data just needs to go to the right place automatically. Figure 19. Physically separated storage tiers force administrators to segregate their traffic and direct it to different types of storage, which is both challenging and frustrating. Drobo Technology Innovation Page 17

18 A Drobo is a single pool of drives by design. The user does not need to think about pooling drives or creating RAID groups. Also, drives of different sizes and types can be used in a Drobo. These advanced technologies make Drobos ideal for allowing SSDs and hard disk drives (HDDs) to work together optimally in the same chassis automatically! To enable use of drives of different sizes, the Drobo virtualization layer (see BeyondRAID section) sits on top of the physical drives. Because Drobos are data aware, data is placed intelligently in the best place on the drives. With the introduction of SSD support in the Drobo B1200i, SSDs provide a faster tier of storage for data that can benefit from faster write and access speeds. Being Data Aware As data enters the Drobo, even while it s still in flight, it s already being examined, but note that performance is not impacted. An additional check has been added to see whether or not the data is transactional in nature. If it is transactional, then the data is written to the faster tier of SSDs. Figure 20. As data is being written to the Drobo, transactional data (shown in red) is written to the faster tier and other data is written to the bulk tier. Leading enterprise storage arrays that offer tiering capabilities can cost more than $50,000, but even then, they are not always aware of what type of data being stored. Often data is first written to so-called bulk storage and then analyzed before it is sent to a high-performance tier and this takes time. With Drobo, transactional data goes directly to SSDs, which ensures that writes AND reads are optimized. Drobo s tiering functionality allows you to store any type of data on the Drobo and the data that can be optimized by SSDs is automatically optimized in flight. This means that any data mixed data types from 10 or more VMs in a vsphere cluster, Exchange databases/datastores, backup data, or files on a file server automatically goes to the right place. Drobo Technology Innovation Page 18

19 One Pool, Multiple Tiers As detailed in the BeyondRAID section, a Drobo is a single pool of drives. Whether you put three drives or 12 drives in a Drobo, they are all part of the same disk pack. Figure 21. Two tiers of storage automatically configured in the same Drobo disk pack. If you have 8 HDDs and 2 SSDs, or 9 HDDs and 3 SSDs, they will be utilized appropriately without any upfront setup. All the administrator has to do is put the drives in the Drobo. Fully Automated Tuning After data is written to storage, data patterns often change. Sometimes a large amount of data is written to storage and not read for weeks or even months. On the other hand, it is not uncommon for a small amount of data to be written and accessed very frequently. Traditional storage arrays treat all data the same way, so if the storage is optimized for streaming performance, then it will most likely be not optimized for transactional performance. Streaming vs. Transactional Performance When streaming data to and from storage, that is, writing and reading files or storing backup data, the data set is large and operations are often sequential. When small bits of data are sent to and from storage, that is, metadata or fields in a database, operations are often random. These two types of data and access patterns require different performance metrics from the storage array. ü When streaming large amounts of data, MB/s stands for how many megabytes per second that can be written to or read from storage is most important. ü When storing data that is transactional in nature, IOPS stands for how many inputs and outputs per second that can be written to or read from storage is most important. When you are evaluating storage to ensure that it can meet the needs of your applications, you need to take both metrics MB/s and IOPS into account. The most intelligent way to have the best of both worlds in business-critical environments is to use storage that has automated tiering capabilities. Drobo Technology Innovation Page 19

20 With Drobo automated tiering, optimization of data doesn t stop when the data reaches a transactional or bulk storage tier. As the stored data is being read off the Drobo, the data is patterned. If data on the bulk tier is frequently read and begins to look more like transactional data, it will be migrated to the transactional tier. If data on the transactional tier becomes cold with very few or no read requests, it will be migrated to the bulk tier. These migrations occur in the background when the storage is not under high load and do not require any administrative interaction. Figure 22. When data becomes cold it is migrated to the bulk tier. When data becomes hot it is migrated to the transactional tier. Tiering Without SSDs Even when a Drobo doesn t contain SSDs, Drobo still puts its automatic tiering feature to work. Without SSDs, the advantage of tiering is to eliminate the write penalty when parity data is created. As discussed in the BeyondRAID section, a Drobo can automatically have zones of different types. Even when all HDDs in the Drobo are the same, Drobo lays out zones differently on the transactional tier than it does on the bulk tier. Figure 23. Drobo creates different zone types on transactional and bulk tiers to tune the storage appropriately for hot and cold data. For example, mirror zones are more optimized for writes and stripes are more optimized for capacity. Leveraging BeyondRAID technology, zones are automatically created and altered on the fly to optimize the Drobo for the type of data you store on it. Drobo Technology Innovation Page 20

21 Just Add SSDs Really! If your disk pack contains only HDDs, they will be optimized automatically, but if your applications need even more performance, all you have to do is introduce at least two SSDs (three for Dual Disk Redundancy) to the Drobo. The SSDs then become the new transactional tier and all of the HDDs are delegated to the bulk tier. This migration occurs automatically, so the only administrative task is to insert new drives into the Drobo. Introducing SSDs into a Drobo increases both write and read performance because SSDs eliminate drive seek time. This performance boost is critical in database and environments in which a lot of small read/write requests are frequent. Industrial Design and User Interface While Drobo s internal firmware capabilities revolutionize the way we use and manage storage, that isn t the only remarkable thing about Drobo. It s unique industrial design and user interface take storage ease-of-use to another level. With traditional storage you must buy drives from the array vendor or you re faced with having to struggle with screw and clips. Then after they re deployed, you discover that they re too large or too loud. When capacity runs low or something goes wrong, you often have to be pro-active and use the admin tool to find out what s going on. Drobo makes this all much easier. Storage Without Screws Every Drobo has a unique carrier-less design, so there is no need to first insert a drive into a carrier. Just insert bare 3.5 drives into a Drobo the same way you would insert cassettes into a tape deck. Figure 24. Bare 3.5 drives are inserted directly into the Drobo. No drive carriers or screws are necessary. Drobo has a latch/spring mechanism, making it easy to insert and eject drives. Just hold down the latch to slide a drive in; and to remove a drive, just press down on the latch and the drive is automatically ejected. Insertion and removal of drives in a Drobo can be done while the Drobo is powered on and in use. It s easy to replace a failed drive and to scale up when you need more capacity without downtime. Drobo Technology Innovation Page 21

22 As Easy as a Traffic Light The primary management interface for a Drobo is the lights on the front of the chassis. Just looking at a Drobo quickly tells you if it s healthy and if enough storage capacity remains. Figure 25. Drive status indicators exist for each drive and a capacity gauge shows what percentage of overall capacity is currently being used. The individual drive status indicators tell you the status of the drives or an action that needs to be taken. For example, if a drive happens to fail, the light for that drive will blink red. If you are running low on storage capacity, an empty drive bay shows yellow, indicating that you should insert another drive in that bay. As you fill your Drobo with data, the capacity gauge starts to illuminate capacity gauge lights, ten in all. For example, if 20% of the overall capacity is used, two lights will be lit. A Drobo can have a capacity gauge because it is data aware (as described in the BeyondRAID section). Figure 26. Drobo is so simple that the instructions fit inside the front cover. They tell you what the lights mean and what action you need to take. Drobo Technology Innovation Page 22

23 Storage Designed for You There are Drobos that are designed for the desk, the desk or rack, or just a rack in a server room/closet. Figure 27. Drobos are available in a wide range of sizes and types to best fit your environment. Unlike most storage arrays, desktop Drobos actually look great on your desk. They are compact and quiet and they look good. Because of the carrier-less architecture, the chassis is not that much larger than the drives it holds. The fan(s) are variable speed and spin fast (make noise) only when the system needs more cooling and spin down when it s not in use. Drobos are carefully designed from the ground up and are made of high-quality materials. One Drobo innovation (becoming more popular in consumer products) is its magnetic front bezel. Instead of being held on by friction clips or tabs, the Drobo front bezel is secured only by magnets. All you have to do is place the bezel on the front of your Drobo and it is automatically aligned and secured. Even in server closets, it s a bezel you will actually want to use! The high-intensity indicator lights shine through the bezel, so whether the bezel is on or off, Drobo s primary form of management (lights) is front and center. Even the lights can be dimmed on some desktop Drobos for comfort in an office or home environment. Advanced Management That s Not Advanced Drobo Dashboard is actually a dashboard. Since the indicator lights on the front are the primary management interface, the number one goal of Dashboard is to show the front of every Drobo in a single window. All directly connected Drobos, as well as Drobos on the local network, automatically appear in Drobo Dashboard for simple, multi-device administration Drobo Technology Innovation Page 23

24 Figure 28. All direct-attached Drobos and networked Drobos automatically appear in Drobo Dashboard. If the lights change on the Drobo, they are also changed in Dashboard. Figure 29. On the left is an actual photo of a Drobo, in the middle is the main Drobo Dashboard view, and the right is the view in the Status pane. Drobo Technology Innovation Page 24

25 Advanced administration is a single click away. Just select the Drobo you wish to manage in Dashboard and navigate to the Advanced pane. Figure 30. Advanced settings pane in Drobo Dashboard. Automatic iscsi Configuration Connecting a server to iscsi storage can be challenging. First, you need to enter the storage IP address, choose a LUN to attach to, provide security credentials (if assigned), and finally format the newly attached volume. With Drobo Dashboard on Windows and Mac OS, attaching to a Drobo iscsi volume is as easy as checking a box. Drobo Technology Innovation Page 25

26 Figure 31. To mount an iscsi volume, just check the box and all host initiator is performed automatically. In Drobo Dashboard, when you mount an iscsi volume, configuration of the iscsi initiator on Windows or Mac is fully automatic, no need for manual configuration. Even CHAP authentication is enforced to the user currently logged in to Drobo Dashboard. Benefit and Cost Savings from Drobo Technology It s clear that Drobo technology makes managing storage and optimizing your applications easier, but this technology also delivers significant cost savings. For most individual professionals and SMB IT directors, capital expenses (CapEx) are the main concern. But in reality operating expenses can outweigh equipment costs, so both need to be considered. All storage arrays that offer drive pooling and redundancy capabilities (like RAID) have these important cost savings: ü Pooling of drives minimizes the number of storage devices that have to be purchased and administered. ü For networked storage, free space is centrally located instead of tied up inside a single server. ü Ability to scale storage means that you don t have to shut down and open up the server. Drobo also gives you the cost savings described in the next few sections. Drobo Technology Innovation Page 26

27 Fewer Drives ü Single pool of storage. All drives in a Drobo are in a single pool of storage, called a disk pack. With Single Disk Redundancy, the space of only one drive is used for parity data, leaving the rest of the drives available for storage. In traditional arrays, there are often multiple RAID groups, each with its own drive used for parity data. ü Thin provisioning. Storage does not have to be provisioned up front, preventing free space from being tied up by a specific computer or application; only the storage needed today is consumed. ü Thin reclamation. When data is deleted on a volume, that freed-up space is returned to the common pool so it that can be used by other volumes for maximum storage efficiency. ü Virtual hot spare. In the event of a drive failure, a Drobo automatically rebuilds itself if enough free space exists, to return to a protected state. With traditional arrays, a hot spare needs to be allocated and designated for this to occur. And that spare drive is just another drive that cannot be used to store data. Fewer Arrays ü Mixed drive utilization. Scale storage seamlessly using larger drives instead of purchasing new arrays to pay as you grow. If you re lucky, by the time you need to add a new drive, the price per TB may be lower than it is today. Lower CapEx ü Carrier-less drive bays. Other storage arrays require drives to be inserted and screwed into a special carrier. Enterprise vendors make you buy the drives from them and pay a carrier tax. Not with Drobo. You can buy the right drives for your application, often saving money along the way. Drobo Technology Innovation Page 27

28 Lower OpEx ü No need to learn storage technologies. You do not need to learn about RAID and pooling of drives, allowing you to spend more time optimizing your environment for applications. ü Automatic tuning. Drobos don t have knobs and switches because everything is automatic. Drobos are application-driven storage instead of being administrator-driven. ü Drive insertion without screws. No messing around with screws to get drives into a Drobo. Drobo s carrier-less design saves time and frustration. ü Simplified management. Drobo s plug-and-play design, elimination of tedious management tasks, and the traffic-light user interface takes storage simplification to another level. Fewer SSDs ü Intelligent use of SSDs. Only a few SSDs are needed in a Drobo. Some large enterprise arrays need 6 or more SSDs per chassis and these drives cost thousands of dollars. Drobo uses SSDs more efficiently than any other array. Summing Up Drobo was created to solve long-term challenges inherent in storage technologies. Those who are familiar with storage assumed RAID was the norm, but that widely used technology, which allows drives to be pooled and protected, could actually evolve. BeyondRAID is a huge advancement for users who do not want to learn about storage and for storage administrators who desire to manage storage less. BeyondRAID, Automated Data-Aware Tiering, and the Drobo User Experience take storage beyond the norm and allow individual professionals and SMB IT access to technologies that they never even knew were available. Drobo will continue to build innovative storage technologies to make the overall experience easier and more powerful. It s not just a focus on storage, but the entire solution that we re focused on in the areas of media, cloud, and virtualization. Drobo Technology Innovation Page 28

29 Appendix A: BeyondRAID vs. Traditional RAID BeyondRAID has significant functional enhancements over traditional RAID options. Drobo Technology Innovation Page 29

30 Appendix B: References Drobo Storage for Business Drobo has created File Sharing (FS) and SAN (iscsi) storage arrays for SMB IT and workgroup environments. Drobo Technology Innovation Page 30

31 Drobo Storage for Professionals Drobo has created Direct-Attached (DAS) and File Sharing (FS) storage arrays for connected home, professional media, and small office environments. Drobo Technology Innovation Page 31

32 Additional Resources Drobo Capacity Calculators: Drobo Demos on Demand: Choosing the Right Drive for Your Drobo: Credits Author: Erik Pounds Contributor: Jason O Broin Editor: Victoria Thomas Cover Graphic: Mike Shin Drobo Technology Innovation Page 32