- Resilient Storage Architecture
- Getting Started
- FAQs
- Initial Data Seed
- HIPAA Compliance
- Product Requirements
- Service Level Agreement
Resilient Storage Architecture
How is it that Symform can offer superior security, reliability, and speed; all at a price
that is much better than the competition? The short answer is Symform's Resilient
Storage ArchitectureTM makes it happen. Symform's founders relied on their experience:
Click to enlarge
Click to enlarge
“When you do the analysis, the decentralized Symform solution provides a very secure and durable
architecture.”
Most cloud storage vendors offer the ability to encrypt data before storing it.
However, encryption is only as good as the strength of the key used. Human
generated keys are too easy to break and too hard to manage. Furthermore,
they make it really hard to de-duplicate data unless it is encrypted with the
same key, which compromises security. Symform does better. We
automatically encrypt each block of data with a unique block key. The key is
a very large 256bit key generated by doing a SHA256 hash on the block itself.
The block is encrypted using federally certified AES-256 encryption algorithm
(Federal minimum is 128 bit encryption – 256 bit encryption is exponentially
stronger). Symform automatically manages these large unique keys as part
of block information in Symform Cloud Control, eliminating both the human
generation and management weak links. If customers are required or desire
to manage their own keys they may encrypt the files before Symform
stores them in the cloud. Symform's protection ensures that even in situations
where customer keys are compromised, the data will still be protected.
Furthermore, by encrypting using the hash of the block as its key, Symform is able to de-duplicate the blocks without needing to decrypt them! If the block already exists in the system, it will not need to be re-uploaded or stored - achieving storage and bandwidth efficiency without compromising any security.
Last but not the least, each encrypted block is subdivided into as many as 96 redundant fragments and these 96 fragments are stored on 96 random machines in different locations. This means that for a perpetrator to compromise the data, they need to first be able to identify these random computers, extract the fragments, and reassemble them. Then they would have to get the randomly generated 256bit key by breaking into Symform's Cloud Control to decrypt the block. They would need to repeat this process for potentially hundreds of blocks just to reassemble one file.
This is much more secure than a typical datacenter based solution, where you might be dealing with human generated keys, key management and all the encrypted data sitting in one location where a single breach compromises everything.
Similarly during recovery, the fragments are brought down in parallel as well to enable 10-50x fast restores. Furthermore, only the fastest 64 pieces are needed to reconstruct the block making the restore process even faster.
Furthermore, Symform's Resilient Storage Architecture enables protecting data in the cloud and creating automatically updated "instant restore" copies available at alternate sites. This completely eliminates the need to even do a restore from the cloud in event of a disaster since data is instantly available at another location.
Each block of encrypted data is divided into as many as 96 pieces and spread to 96 completely independent computers located in different geographies. Any 64 out of the 96 pieces are needed to reconstruct the block. In other words 33 unrelated, unknown, and geographically dispersed disks would have to fail at nearly the same time before any customer data is lost. This ensures that even a widespread regional disaster is highly unlikely to result in a data loss. Compare this to a typical datacenter approach where typically RAID5 or RAID6 storage is used where just two or three simultaneous disk failures inside a single computer will result in data loss. Symform's Resilient Storage Architecture inherently avoids these failures. Aggregating resources over a geographic spread eliminates multiple correlated failures due to hardware, infrastructure or humans that are common inside data centers. The chances that 33 completely independent computers located in completely different geographies that happen to have pieces of the same block of data will fail at the same time is very, very, very, very… small.
But that's not good enough for Symform. We know machines fail. Unlike centralized datacenters, Symform does not depend on human processes to deal with these failures. Symform's Cloud Control constantly monitors all the devices in the Storage Cloud. When it detects that a particular computer has failed, it automatically triggers regeneration of the fragments that were stored on that computer and repopulates them elsewhere in the network. The system prevents data loss by self-healing automatically.
The result is lifelong durability of your data and ability to survive any kind of disaster – from something as simple as fat-finger delete of a file or spilled coffee on the server to something more devastating such as fire, theft or natural disaster.
Is this a fair comparison? Why aren't we comparing Symform with the much more expensive storage solutions from EMC? The reason is simple. Cloud Storage is not for applications needing high performance and low latency storage. Cloud Storage is "far away" storage. It is great for backup, archival and file sharing type applications where medium performance and latency is acceptable along with some of its other benefits such as offsite protection, ubiquitous access, ability to grow/shrink on demand, etc. Today businesses use commodity storage solutions like tapes, disks, and file servers to accomplish this.
For cloud storage to gain mainstream adoption, it must fit into existing IT budgets for commodity storage. As it turns out, only a small percentage of the data businesses need to store and manage is mission critical and needs high performance, low latency throughput. The majority of business data is stored in commodity storage solutions and can move to Cloud Storage. This is also the data that is seeing maximum growth.
The underlying reason for the outrageous price difference between cloud and local storage is the cost of building and operating data centers that are used to provide cloud storage services by other vendors. With all the hype around cloud computing and big companies like Microsoft, Amazon and Google spending billions of dollars building data centers to offer cloud computing services, it is highly counter intuitive to think that this approach is broken. Most cloud computing vendors argue significant economies of scale achieved by building mammoth data centers to offer cloud services.
There are indeed some economies of scale in data centers. These vendors are able to use their buying power to get 20-30% discounts on buying hardware and power in volume. Unfortunately, these mammoth data centers also come with significant diseconomies of scale. For example, power represents one of the most significant costs in data centers. Buying power at volume requires getting it on high voltage lines. This high voltage power needs to be stepped down and distributed to the thousands of DC powered servers and networking gear. On average 50% power is lost in step down and distribution. This evaporates the economies of scale. What is worse is that putting thousands of servers together requires active cooling infrastructure, which consumes even more power!
Another example is utilization. A Gartner study found that most datacenters are on average only 15% utilized. This is because datacenters need to be provisioned for peak usage even though peak usage only occurs a few days in a given year. This means that costs of unutilized 80% resources burdens the 20% getting utilized – creating a 5x cost differential.
In stark contrast, Symform found that most servers and PCs inside businesses end up getting recycled with 50% of their disk space unused. Many even have enough horsepower to add one or more additional commodity multi-TB hard drives if needed. However, most of this capacity was going to waste because there wasn't any technology to smartly aggregate it to create a virtual storage pool.
Symform's Resilient Storage Architecture harnesses this existing unused capacity inside businesses and turns it into valuable Cloud Storage. By doing so, not only are we maximally utilizing resources that are already paid for, we are also eliminating significant additional data center costs including power, real estate, bandwidth, operations, etc. that businesses would have to double pay for to get the benefits of cloud storage.
Symform's Resilient Storage Architecture enabled creation of a public storage pool over the Internet for a large number of businesses of any size to join and get the benefits of Internet scale storage pool. Large corporations with global presence can also choose to get a private storage pool built where they tap their own unused resources across their global corporate network and still get all the benefits of the architecture as well as control the physical hardware underneath.
Just to be clear, not all uses of data centers are inappropriate. Data centers are the only way to solve certain application needs – e.g. high transaction database applications where the data and application logic needs to be in close proximity for high performance and low latency throughput. Even Symform uses data centers for the smart brain that orchestrates the Symform Storage Cloud system. We call it the Symform Cloud Control. However, the overall data that the Symform Storage Cloud stores is 4 orders of magnitude more than what is stored and used by the Cloud Control. In other words, the amount of data we are storing in a data center is 0.01% of the overall data in the system. Oh, by the way, Cloud Control is also mirrored and geo-distributed.
That's the long version of how Symform's Resilient Storage Architecture is truly able to provide valuable Cloud Storage at the price of local commodity storage.
If you are interested in protecting your data now with Symform please click here to get your 200GB free Symform account. Or contact us for more information.
- Securing the world's largest distributed installed base, and
- Building enterprise datacenters from commodity hardware
Exchange Cheap Local Storage for More Valuable Cloud Storage
Click to enlarge
Encrypted. Shredded. Geo-Spread. Self Healing.
Click to enlarge
Impermeable Protection
“When you do the analysis, the decentralized Symform solution provides a very secure and durable
architecture.”
- Ginny Roth, Lab Engineer and Analyst, ESG
Furthermore, by encrypting using the hash of the block as its key, Symform is able to de-duplicate the blocks without needing to decrypt them! If the block already exists in the system, it will not need to be re-uploaded or stored - achieving storage and bandwidth efficiency without compromising any security.
Last but not the least, each encrypted block is subdivided into as many as 96 redundant fragments and these 96 fragments are stored on 96 random machines in different locations. This means that for a perpetrator to compromise the data, they need to first be able to identify these random computers, extract the fragments, and reassemble them. Then they would have to get the randomly generated 256bit key by breaking into Symform's Cloud Control to decrypt the block. They would need to repeat this process for potentially hundreds of blocks just to reassemble one file.
This is much more secure than a typical datacenter based solution, where you might be dealing with human generated keys, key management and all the encrypted data sitting in one location where a single breach compromises everything.
Uncompromising Cloud Speed
The 96 pieces of a block are distributed to 96 different computers over parallel Internet connections resulting in as much as 10-50x faster speed vs. sending all the data to one server inside a data center, which is likely sharing bandwidth with hundreds or thousands of other servers. This makes uploading data to Symform Storage Cloud faster than any other alternative in the market.Similarly during recovery, the fragments are brought down in parallel as well to enable 10-50x fast restores. Furthermore, only the fastest 64 pieces are needed to reconstruct the block making the restore process even faster.
Furthermore, Symform's Resilient Storage Architecture enables protecting data in the cloud and creating automatically updated "instant restore" copies available at alternate sites. This completely eliminates the need to even do a restore from the cloud in event of a disaster since data is instantly available at another location.
Disaster-Proof Durability
We all know that servers and hard disks fail. We also know that humans make catastrophic mistakes. After all, that is why we want to use cloud storage to protect our valuable data. Well, turns out computers and hard disks fail in datacenters as well. Human failures are common too. Given the concentration of IT resources, the potential for correlated failures and errors is significantly higher with a centralized datacenter. Furthermore, the concentration of heat generating hardware creates a higher than normal fire hazard, which is why datacenters often talk about multiple types of fire suppression.Each block of encrypted data is divided into as many as 96 pieces and spread to 96 completely independent computers located in different geographies. Any 64 out of the 96 pieces are needed to reconstruct the block. In other words 33 unrelated, unknown, and geographically dispersed disks would have to fail at nearly the same time before any customer data is lost. This ensures that even a widespread regional disaster is highly unlikely to result in a data loss. Compare this to a typical datacenter approach where typically RAID5 or RAID6 storage is used where just two or three simultaneous disk failures inside a single computer will result in data loss. Symform's Resilient Storage Architecture inherently avoids these failures. Aggregating resources over a geographic spread eliminates multiple correlated failures due to hardware, infrastructure or humans that are common inside data centers. The chances that 33 completely independent computers located in completely different geographies that happen to have pieces of the same block of data will fail at the same time is very, very, very, very… small.
But that's not good enough for Symform. We know machines fail. Unlike centralized datacenters, Symform does not depend on human processes to deal with these failures. Symform's Cloud Control constantly monitors all the devices in the Storage Cloud. When it detects that a particular computer has failed, it automatically triggers regeneration of the fragments that were stored on that computer and repopulates them elsewhere in the network. The system prevents data loss by self-healing automatically.
The result is lifelong durability of your data and ability to survive any kind of disaster – from something as simple as fat-finger delete of a file or spilled coffee on the server to something more devastating such as fire, theft or natural disaster.
Unbeatable Savings
You know that the traditional Cloud Storage model is broken when you can buy 2TB drive for less than $100 but to back up 2TB of business data costs $12,000/year with MozyPro and much more at some of the other providers. And, local storage is getting cheaper all the time so the gap between local and cloud is actually growing!Is this a fair comparison? Why aren't we comparing Symform with the much more expensive storage solutions from EMC? The reason is simple. Cloud Storage is not for applications needing high performance and low latency storage. Cloud Storage is "far away" storage. It is great for backup, archival and file sharing type applications where medium performance and latency is acceptable along with some of its other benefits such as offsite protection, ubiquitous access, ability to grow/shrink on demand, etc. Today businesses use commodity storage solutions like tapes, disks, and file servers to accomplish this.
For cloud storage to gain mainstream adoption, it must fit into existing IT budgets for commodity storage. As it turns out, only a small percentage of the data businesses need to store and manage is mission critical and needs high performance, low latency throughput. The majority of business data is stored in commodity storage solutions and can move to Cloud Storage. This is also the data that is seeing maximum growth.
The underlying reason for the outrageous price difference between cloud and local storage is the cost of building and operating data centers that are used to provide cloud storage services by other vendors. With all the hype around cloud computing and big companies like Microsoft, Amazon and Google spending billions of dollars building data centers to offer cloud computing services, it is highly counter intuitive to think that this approach is broken. Most cloud computing vendors argue significant economies of scale achieved by building mammoth data centers to offer cloud services.
There are indeed some economies of scale in data centers. These vendors are able to use their buying power to get 20-30% discounts on buying hardware and power in volume. Unfortunately, these mammoth data centers also come with significant diseconomies of scale. For example, power represents one of the most significant costs in data centers. Buying power at volume requires getting it on high voltage lines. This high voltage power needs to be stepped down and distributed to the thousands of DC powered servers and networking gear. On average 50% power is lost in step down and distribution. This evaporates the economies of scale. What is worse is that putting thousands of servers together requires active cooling infrastructure, which consumes even more power!
Another example is utilization. A Gartner study found that most datacenters are on average only 15% utilized. This is because datacenters need to be provisioned for peak usage even though peak usage only occurs a few days in a given year. This means that costs of unutilized 80% resources burdens the 20% getting utilized – creating a 5x cost differential.
In stark contrast, Symform found that most servers and PCs inside businesses end up getting recycled with 50% of their disk space unused. Many even have enough horsepower to add one or more additional commodity multi-TB hard drives if needed. However, most of this capacity was going to waste because there wasn't any technology to smartly aggregate it to create a virtual storage pool.
Symform's Resilient Storage Architecture harnesses this existing unused capacity inside businesses and turns it into valuable Cloud Storage. By doing so, not only are we maximally utilizing resources that are already paid for, we are also eliminating significant additional data center costs including power, real estate, bandwidth, operations, etc. that businesses would have to double pay for to get the benefits of cloud storage.
Symform's Resilient Storage Architecture enabled creation of a public storage pool over the Internet for a large number of businesses of any size to join and get the benefits of Internet scale storage pool. Large corporations with global presence can also choose to get a private storage pool built where they tap their own unused resources across their global corporate network and still get all the benefits of the architecture as well as control the physical hardware underneath.
Just to be clear, not all uses of data centers are inappropriate. Data centers are the only way to solve certain application needs – e.g. high transaction database applications where the data and application logic needs to be in close proximity for high performance and low latency throughput. Even Symform uses data centers for the smart brain that orchestrates the Symform Storage Cloud system. We call it the Symform Cloud Control. However, the overall data that the Symform Storage Cloud stores is 4 orders of magnitude more than what is stored and used by the Cloud Control. In other words, the amount of data we are storing in a data center is 0.01% of the overall data in the system. Oh, by the way, Cloud Control is also mirrored and geo-distributed.
That's the long version of how Symform's Resilient Storage Architecture is truly able to provide valuable Cloud Storage at the price of local commodity storage.
If you are interested in protecting your data now with Symform please click here to get your 200GB free Symform account. Or contact us for more information.
