Amazon AWS and Cloudberry Backup
Being in IT for over 20 years, having been there, done that, I have seen just about everything under the sun in the IT industry, so it takes quite a bit to get me to 'geek out' and get excited about a solution, but Amazon AWS and Cloudberry Backup together create a very slick solution. Previously, I used GreenGeeks for my web site hosting, and Crashplan for my backup service. I had no issues with either service, but they were both coming up for renewal and since I had pre-paid for both services years in advance, I was losing the associated discounts, and both services would revert to $10 per month each. Granted, that is not a lot in the enterprise world, but there is no sense paying more than necessary especially for personal and not professional use. I run a Windows 2016 server at home for my home network- it controls Active Directory authentication for our family, DNS, DHCP, and is attached to a Drobo redundant storage device. There are critical documents and family archives that exist on the storage array and on the server, and thus I feel that it is critical to maintain a proper disaster recovery plan involving on-site and off-site storage and as much redundancy as is practicable.
I have designed and implemented many DR plans over the years for many hundred-million dollar plus companies. But whether you're designing a DR plan at the enterprise level or at the home level, the principles are the same. You need to plan for a complete loss of hardware and data, in a worst case scenario, and determine the acceptable down-time associated with such a loss. Obviously, it makes sense to plan in as much redundancy as you can afford based on your needs as part of your basic strategy to minimize or eliminate down-time where possible. Along those lines, I have redundant Internet connections from disparate Internet Service Providers (ISPs) with real-time failover at the firewall, so my network automatically switches ISPs in the event that one fails. My server has redundant power supplies, and redundant drives in a RAID5 array. Data such as software, music, videos, photos, and other critical documents reside in an attached Drobo storage system which contains redundant drives as well, so in the event that a drive fails, data is not lost. I use folder redirection so that user's documents are stored on the server and not on individual PCs. All of those devices are protected behind a quality sine-wave UPS, and our home has an additional 20kWh battery backup system with solar array and generator input. Physical security is another concern, for which we have multiple redundant layers in place, but I won't go into the details of that for obvious reasons. Those are the kind of things you need to think through for resistance to normal operation interruptions such as loss of Internet, power, drives, etc. Beyond that, the possibility always exists for complete hardware and data loss, in the event of an extreme event such as a tornado or fire or just catastrophic hardware failure. That is when backups come into play and pay for themselves. Backups are like insurance, you hope you never need them, but when you do, boy are you glad you have them. There are two main components to my backup plan - Amazon AWS (Amazon Web Services) and Cloudberry Backup.
First, Amazon has some of the largest data centers in multiple geographic locations around the world, offering an appealing destination for data and a pay for only what you use billing system. Their S3 "simple storage service" is a great option for many people and many large companies such as Netflix. I am utilizing it in multiple ways. I am now hosting my web site on Amazon S3 essentially for pennies per month. I am also moving about 400GB of family home movies to Amazon S3 "IA" or infrequent access storage, thus providing an off-site home for that critical data allowing me to exclude it from my regular backup plan. The S3IA storage level is perfect for infrequently accessed data, as it costs about half the amount that S3 is, but adds a small retrieval fee per GB when the data is accessed, which in my case is rarely, but it still needs to be available quickly for when it is accessed. Additionally, I'm using Amazon Glacier storage as the destination for my backup data including the images from my server and the critical data on my Drobo. Glacier is only .4 cents per GB per month, so very inexpensive storage, yet has the same level of redundancy and data durability as other Amazon storage methods, and, just like S3 storage, is spread among multiple Amazon availability zones, in the event that they have a data center failure. Glacier, however, is slower to access, meaning that standard retrieval requests (currently 1 cent per GB) are served up in 3-5 hours, or you can make an expedited retrieval request for 3 cents per GB and receive your data in just a few minutes. While Amazon is quiet about the technology behind this system, I believe Glacier is based on a very large robotic tape library, which would explain the low data storage cost and the slow standard retrieval time frames - a perfect solution for 'cold' data storage. I am also using Amazon's route 53 DNS hosting service. Amazon also has a service called EC2 where you can easily spin up servers on their infrastructure. During your first year, you can even run a small server for free on Amazon within their 'free tier.' There are now dozens of AWS services available for hosting servers, data, and even executing code directly- making AWS a very appealing platform for many web services.
Second, while Amazon offers a great cloud data destination for these backups, you need a way to actually backup your data to Amazon. Hands down, the best solution I have found to do this during my research was Cloudberry Backup. It provides advanced capabilities that are on-par with enterprise services I work with in my professional experience that are far more costly. I have two backups setup within Cloudberry Backup on my server - one is a real-time data backup which monitors our users' documents and critical data on my Drobo and uploads those changes to Amazon Glacier in-real time within seconds after a file is changed. Cloudberry also supports powerful versioning rules and ransomware protection to ensure that in the event your files are mass-encrypted (as in the case of a ransomware infection) or mass-deleted, your data is still safe- which is a critically important feature. The second backup I have setup is an image backup of my server C drive (operating system) and E drive (userdata). This backup can be made to external media, such as a USB drive, or directly to the cloud, or both! In my case, I am using Cloudberry's "hybrid backup" feature that runs the block-level image backup to a USB hard drive, and then backs that same data up to a cloud destination, in my case Amazon S3 (which then immediately converts it to Glacier storage via an S3 rule). Both backups- local and cloud can be used for bare-metal system recovery, which is a powerful (and important) option. In the event of a major server failure, once your hardware is replaced, you simply boot it up using Cloudberry's USB or CD boot disk, and either plug in the USB hard drive, if available, or connect to your cloud backup, and restore directly to the new server. Additionally, if you're in a time crunch waiting on parts to repair your server, you can actually spin up the backed up server images directly in the cloud on Amazon EC2 or Microsoft Azure- with no hardware at all! You can then connect your server running in the cloud to your network using an IPSEC VPN tunnel. Those are certainly enterprise grade DR capabilities and it's great that those options are now available from a reasonably priced software package. Thus, I have added the capability of image-based restoration locally and in the cloud to my DR plan, while saving money in the long run.
In the end, I have calculated that my Amazon bill will total only about $6 per month, as compared to the $20 per month I would have spent with Greengeeks and Crashplan, all thanks to Amazon AWS and Cloudberry Backup. If you are interested in either, I recommend that you check them out, Amazon AWS offers a plethora of free services for a year to use and experiment with, and Cloudbery Backup offers a 15 day fully functional free trial.you can download at http://www.cloudberrylab.com/best-bare-metal-backup-software.aspx to see if it works for you.
I have designed and implemented many DR plans over the years for many hundred-million dollar plus companies. But whether you're designing a DR plan at the enterprise level or at the home level, the principles are the same. You need to plan for a complete loss of hardware and data, in a worst case scenario, and determine the acceptable down-time associated with such a loss. Obviously, it makes sense to plan in as much redundancy as you can afford based on your needs as part of your basic strategy to minimize or eliminate down-time where possible. Along those lines, I have redundant Internet connections from disparate Internet Service Providers (ISPs) with real-time failover at the firewall, so my network automatically switches ISPs in the event that one fails. My server has redundant power supplies, and redundant drives in a RAID5 array. Data such as software, music, videos, photos, and other critical documents reside in an attached Drobo storage system which contains redundant drives as well, so in the event that a drive fails, data is not lost. I use folder redirection so that user's documents are stored on the server and not on individual PCs. All of those devices are protected behind a quality sine-wave UPS, and our home has an additional 20kWh battery backup system with solar array and generator input. Physical security is another concern, for which we have multiple redundant layers in place, but I won't go into the details of that for obvious reasons. Those are the kind of things you need to think through for resistance to normal operation interruptions such as loss of Internet, power, drives, etc. Beyond that, the possibility always exists for complete hardware and data loss, in the event of an extreme event such as a tornado or fire or just catastrophic hardware failure. That is when backups come into play and pay for themselves. Backups are like insurance, you hope you never need them, but when you do, boy are you glad you have them. There are two main components to my backup plan - Amazon AWS (Amazon Web Services) and Cloudberry Backup.
First, Amazon has some of the largest data centers in multiple geographic locations around the world, offering an appealing destination for data and a pay for only what you use billing system. Their S3 "simple storage service" is a great option for many people and many large companies such as Netflix. I am utilizing it in multiple ways. I am now hosting my web site on Amazon S3 essentially for pennies per month. I am also moving about 400GB of family home movies to Amazon S3 "IA" or infrequent access storage, thus providing an off-site home for that critical data allowing me to exclude it from my regular backup plan. The S3IA storage level is perfect for infrequently accessed data, as it costs about half the amount that S3 is, but adds a small retrieval fee per GB when the data is accessed, which in my case is rarely, but it still needs to be available quickly for when it is accessed. Additionally, I'm using Amazon Glacier storage as the destination for my backup data including the images from my server and the critical data on my Drobo. Glacier is only .4 cents per GB per month, so very inexpensive storage, yet has the same level of redundancy and data durability as other Amazon storage methods, and, just like S3 storage, is spread among multiple Amazon availability zones, in the event that they have a data center failure. Glacier, however, is slower to access, meaning that standard retrieval requests (currently 1 cent per GB) are served up in 3-5 hours, or you can make an expedited retrieval request for 3 cents per GB and receive your data in just a few minutes. While Amazon is quiet about the technology behind this system, I believe Glacier is based on a very large robotic tape library, which would explain the low data storage cost and the slow standard retrieval time frames - a perfect solution for 'cold' data storage. I am also using Amazon's route 53 DNS hosting service. Amazon also has a service called EC2 where you can easily spin up servers on their infrastructure. During your first year, you can even run a small server for free on Amazon within their 'free tier.' There are now dozens of AWS services available for hosting servers, data, and even executing code directly- making AWS a very appealing platform for many web services.
Second, while Amazon offers a great cloud data destination for these backups, you need a way to actually backup your data to Amazon. Hands down, the best solution I have found to do this during my research was Cloudberry Backup. It provides advanced capabilities that are on-par with enterprise services I work with in my professional experience that are far more costly. I have two backups setup within Cloudberry Backup on my server - one is a real-time data backup which monitors our users' documents and critical data on my Drobo and uploads those changes to Amazon Glacier in-real time within seconds after a file is changed. Cloudberry also supports powerful versioning rules and ransomware protection to ensure that in the event your files are mass-encrypted (as in the case of a ransomware infection) or mass-deleted, your data is still safe- which is a critically important feature. The second backup I have setup is an image backup of my server C drive (operating system) and E drive (userdata). This backup can be made to external media, such as a USB drive, or directly to the cloud, or both! In my case, I am using Cloudberry's "hybrid backup" feature that runs the block-level image backup to a USB hard drive, and then backs that same data up to a cloud destination, in my case Amazon S3 (which then immediately converts it to Glacier storage via an S3 rule). Both backups- local and cloud can be used for bare-metal system recovery, which is a powerful (and important) option. In the event of a major server failure, once your hardware is replaced, you simply boot it up using Cloudberry's USB or CD boot disk, and either plug in the USB hard drive, if available, or connect to your cloud backup, and restore directly to the new server. Additionally, if you're in a time crunch waiting on parts to repair your server, you can actually spin up the backed up server images directly in the cloud on Amazon EC2 or Microsoft Azure- with no hardware at all! You can then connect your server running in the cloud to your network using an IPSEC VPN tunnel. Those are certainly enterprise grade DR capabilities and it's great that those options are now available from a reasonably priced software package. Thus, I have added the capability of image-based restoration locally and in the cloud to my DR plan, while saving money in the long run.
In the end, I have calculated that my Amazon bill will total only about $6 per month, as compared to the $20 per month I would have spent with Greengeeks and Crashplan, all thanks to Amazon AWS and Cloudberry Backup. If you are interested in either, I recommend that you check them out, Amazon AWS offers a plethora of free services for a year to use and experiment with, and Cloudbery Backup offers a 15 day fully functional free trial.you can download at http://www.cloudberrylab.com/best-bare-metal-backup-software.aspx to see if it works for you.
Comments