How to minimize backup costs

Table of Contents

Case
Solution

Case #

There are cases in which you need to estimate the overall costs for a backup solution, including both on-premises and cloud components. In each case you need to minimize backup costs to achieve overall cost optimization. Ideally, in the context of a well-designed business continuity plan (BCP), you need to have both on-site and off-site backups at all times. This article provides guidance on how to minimize backup costs by discussing the types of backup workloads and some notable backup tools and solutions for each workload which can potentially minimize overall backup costs.

Solution #

Backup workloads and storage size calculation #

First off, you need to determine the workloads you need to backup as well as the estimated storage sizes for each workload. Refer to another KB article about how to estimate your backup storage sizes.

The following list presents the most common backup workloads, assume the backups are taken primarily from on-premise to on-premise, from on-premise to cloud or from cloud to another cloud.

Server bare metal backups
Operating system and Active Directory state backups
File data
Active Directory virtual machines (application-aware backup)
SQL database data (application-aware backup)
Non-SQL database data
Virtual machine data
Container data
Mailbox data
Cloud-to-cloud backup scenarios

Backup storage types and methods #

The following are the most prominent backup storage types, which can be used as target storage for your backups.

File storage. Common protocols used for file storage are SMB, CIFS and NFS.
Block storage. Common protocols used for block storage are iSCSI, FC and FCoE. Also there are public cloud implementations of block storage such as Azure Managed Disk. Azure managed disks are block-level storage volumes that are managed by Azure and used with Azure Virtual Machines.
Object storage. Amazon S3 and Azure blob storage are types of object storage implementation. There are also numerous S3-compliant object storage services, including the Backblaze B2 and the Synology C2 object storage services.
Data lake storage, which is a special type of blob storage used in big data analytics. An example of such a storage service is Azure Data Lake.

The fundamental backup methods are the following:

Backup as a service (BaaS). An example of such as a solution is the Azure Backup cloud service.
Backup software utilizing your own storage as target. A common solution for local storage are NAS devices, which are more cost effective than their SAN equivalents. A good overall solution is the Synology NAS family, which offers various native apps for local and cloud backup. In the case of Synology NAS, you should first choose the appropriate NAS appliance based on your requirements and needed features. Then you can make use of the free NAS “Active Backup for Business” application for local NAS backups (https://www.synology.com/en-global/dsm/feature/active-backup-business/overview), “the Active Backup for M365” application for local NAS backups and the “Hyper Backup” application for taking backups to Backblaze or other cloud storage (https://www.backblaze.com/docs/cloud-storage-integrate-synology-hyper-backup-with-backblaze-b2). Also SQL server instances can be backed up by following the method listed in the following article: https://kb.synology.com/en-uk/DSM/tutorial/back_up_and_restore_MS_SQL_Server_via_ABB. There are certain technical requirements and limitations (https://www.synology.com/en-global/dsm/7.2/software_spec/abb) which must be taken into account.
Backup software utilizing cloud storage as target

Backup cost metrics #

Each storage vendor has different metrics upon which the final backup costs will be incurred, including the following metrics, which must be minimized.

Backup software, backup server and backup agents. In case you have your own backup software, this cost will be eliminated. In a Backup as a Service (BaaS) solution, this cost is included in a subscription-based billing model. Also when utilizing a backup software solution, most likely you will need to account for a dedicated or collocated backup server (running on Windows or Linux) and you may also need to install backup agents in all machines to be backed up. Read on to find out more about free or cost effective backup software solutions.
Total network throughput and number of transactions occurred during a monthly time frame. A transaction refers to any time an API call is made with your account, such as listing the contents of your storage bucket or downloading a file. Transactions can be classified as class A, class B or class transactions. An explanation of the types of API calls group in each class of transactions in the example of the Backblaze cloud service can be found at https://www.backblaze.com/b2/b2-transactions-price.html.
Total upload and download bandwidth. Usually the download bandwidth (egress for the cloud storage vendor) is charged separately, while the upload costs (ingress for the cloud storage vendor) has lower or zero cost.
Churn. "Churn" refers to "% change in the data being backed up between two consecutive backups". This could be due to addition of new data, modification or deletion of existing data. Since Azure Backup only backs up what changes between two backups (forever incremental), this value governs the rate of growth of the backup storage. If there is no churn, no backup data is sent to the vault and therefore backup storage doesn't grow. But if there is say an average 5% daily churn, the backup storage growth will be 5% with every daily backup. Typically File workloads have low churn, VMs have moderate churn while active DBs like SQL have larger churn.
Backup storage size. Refer to another KB article about how to estimate your backup storage sizes. Opex online disks are most often more cost effective than buying and maintaining Capex disks on-site.