Moving to the cloud doesn’t mean choosing between “everything in our server room” and “everything in AWS”. Some systems can stay in a private cloud, others can move to a public one, and email and documents can go to SaaS services such as Microsoft 365. If these parts are connected and managed as one whole, that is a hybrid cloud.
Below: what a hybrid cloud is, when it makes sense, what it takes to make it work, and common mistakes.
What a hybrid cloud is
NIST defines a hybrid cloud as a composition of two or more distinct cloud infrastructures — private, community or public. Each remains a separate entity, but they are bound together by technology that allows data and applications to move between them.
In practice, for a business this usually means:
- a private cloud — your own or rented servers combined into a managed virtualisation platform (more on private clouds);
- a public cloud — virtual servers, databases and storage in AWS or Microsoft Azure;
- SaaS — ready-made services, such as email and documents in Microsoft 365.
Two sites on their own are not yet a hybrid. What makes a setup hybrid is the links between them: a shared network, a single set of user accounts, shared monitoring and the ability to move data and workloads. How the IaaS, PaaS and SaaS models differ — see “IaaS, PaaS and SaaS: what to choose for business”.
When a hybrid makes sense
| Situation | What goes where |
|---|---|
| Regulated data or requirements on where data is stored | regulated systems in a private cloud or with a provider that meets the requirements; the rest in a public cloud |
| A steady round-the-clock workload | work out where it is cheaper: on your own hardware or in a public cloud with a commitment-based price |
| A seasonal or variable workload | in a public cloud, where resources can be added and released quickly |
| Recently purchased hardware | keep systems on it until it has served its term |
| Systems tied to hardware in the office or on the shop floor | on premises, connected to the cloud |
| A standby site is needed | main systems in a private cloud and the standby site in a public one, or the other way round |
| A gradual migration | hybrid as a transition stage: systems move in waves, and old and new run side by side |
For financial institutions, requirements for cloud services are set by the NBU — see “NBU Resolution 99: cloud for banks and fintech”. How to estimate and compare costs — see “Cloud costs: how to estimate and not overpay”.
What it takes to make a hybrid work
Network
There are two main ways to connect the sites:
- Site-to-site VPN. An encrypted tunnel over the internet. This is how VPN Gateway works in Azure, for example. No separate line is needed, but the connection depends on the quality of your internet link.
- A private connection. In Azure this is ExpressRoute, in AWS — Direct Connect. The connection is set up through a connectivity provider, and traffic doesn’t go over the public internet. Microsoft notes that such connections offer more reliability, faster speeds and more consistent latency. A VPN can stay as a backup path.
Plan IP addressing in advance so that the address ranges in the private and public clouds don’t overlap. Each site needs protection and segmentation at its edge — see network security and NGFW. And remember the cost: outbound traffic from a public cloud and VPN gateways are billed.
User accounts
Users should sign in to on-premises and cloud systems with the same credentials. Microsoft calls this hybrid identity. Accounts from on-premises Active Directory are synchronised to Microsoft Entra ID, and there are three ways to sign in:
- password hash synchronisation — the password is checked in the cloud, no extra infrastructure is needed;
- pass-through authentication — on-premises Active Directory checks the password through agents on your servers;
- federation — sign-in through a separate system, such as AD FS.
Microsoft recommends turning on password hash synchronisation with any of these methods as a backup. If the on-premises infrastructure goes down, for example after a ransomware attack, sign-in can be switched over to it — and users can work with cloud services and email again. The switch isn’t automatic, so describe it in your recovery plan. How to protect Active Directory itself — see “Active Directory security: the basic settings”.
Security and monitoring
The same rules for access, updates and logging in both environments. Logs from the private and the public cloud should be collected in one place; otherwise an attack that started in one environment and moved to the other will be hard to spot. Remember the shared responsibility model: data, accounts and access are yours to look after, wherever a system runs.
Backups
A hybrid gives you a convenient second site: copies of on-premises systems can be stored in the public cloud, and copies of cloud data in the private one. The 3-2-1 rule still applies — see “3-2-1 backup rule: protecting business data” and “Disaster recovery: RTO, RPO and a recovery plan”.
Management
Keep a single list: which system runs where, who owns it and what its availability requirements are. Without one, a hybrid quickly turns into a collection of unrelated sites.
Common mistakes
- Cloud resources appear “ad hoc”, with no network plan, owners or cost control.
- The sites’ IP ranges overlap, and they can’t be connected without address translation.
- Separate user accounts in each environment.
- The cost of outbound traffic and connectivity isn’t taken into account.
- Backups are kept on the same site as the data.
- There is no shared security monitoring.
Checklist
- For each system it is decided where it runs and why
- Regulated data is placed according to the requirements
- A way to connect the sites is chosen, and there is a backup link
- IP address ranges don’t overlap
- There is an NGFW and segmentation at the edge of each site
- User accounts are synchronised, multi-factor authentication is on
- Logs from both environments are collected and analysed
- Backups are stored on another site, and restores are tested
- Traffic and connectivity costs are in the budget
How Bereg-IT can help
We build and support private clouds and help choose a public platform. We combine them in a hybrid setup: we decide what goes where, design the connection between the sites and network protection with an NGFW, set up backups between the sites and move systems according to a plan. SOC as a Service collects events from both environments. You can start with a consultation or an audit; the move itself is described in “Cloud migration: a step-by-step plan for business”.
The first consultation is free. Send a request and we’ll suggest the setup that fits your systems.
Sources
- NIST SP 800-145, The NIST Definition of Cloud Computing (2011) — definition of a hybrid cloud: https://csrc.nist.gov/pubs/sp/800/145/final
- Microsoft Learn, About Azure VPN Gateway — site-to-site VPN, pricing: https://learn.microsoft.com/en-us/azure/vpn-gateway/vpn-gateway-about-vpngateways
- Microsoft Learn, Azure ExpressRoute Overview — a private connection to the cloud: https://learn.microsoft.com/en-us/azure/expressroute/expressroute-introduction
- AWS, What is Direct Connect? — a private connection to AWS: https://docs.aws.amazon.com/directconnect/latest/UserGuide/Welcome.html
- Microsoft Learn, What is hybrid identity with Microsoft Entra ID?: https://learn.microsoft.com/en-us/entra/identity/hybrid/whatis-hybrid-identity
- Microsoft Learn, Authentication for Microsoft Entra hybrid identity solutions — sign-in methods and the recommendation on password hash synchronisation: https://learn.microsoft.com/en-us/entra/identity/hybrid/connect/choose-ad-authn
- Microsoft Learn, Shared responsibility in the cloud — who is responsible for what: https://learn.microsoft.com/en-us/azure/security/fundamentals/shared-responsibility