Cloud Computing: What Is It? Beginner’s Guide 2026

[Published: 2026-07-05 | Last updated: 2026-07-05]

TL;DR

• Cloud computing is the delivery of storage, servers, databases, software, and other IT services over the internet instead of on a local computer or office server.
• The cloud usually follows a pay-as-you-go model, which means you pay for what you use instead of buying and maintaining all the hardware upfront.
• Public cloud, private cloud, hybrid cloud, and multi-cloud are the main deployment models, and each one fits a different mix of cost, control, and compliance needs.
• Major cloud providers include Amazon Web Services (AWS), Microsoft Azure, and Google Cloud, which together accounted for a large share of enterprise cloud use in Gartner market coverage for 2024 (Gartner, 2024).
• If you want a plain-English companion on the software side, our computer software guide is a useful follow-up.

What Is Cloud Computing, and Why It Matters in 2026?

Cloud computing is internet-delivered computing power, storage, and software that you use on demand instead of running everything on your own device. For most people, that means apps, files, and services live on remote systems, while your phone, tablet, or laptop is just the access point.

That shift matters because it changes who owns the hardware burden. A person or business can use more computing power without buying a server room, which is why cloud computing now sits under email, streaming, file sync, online collaboration, and much of modern AI.

[IMAGE: A simple diagram showing a laptop connecting through the internet to cloud servers, storage, and applications]

Cloud computing is often compared to electricity. You do not build your own power plant to turn on a lamp, you plug into the grid. Cloud services work the same way, except the utility is computing power instead of electricity.

For a plain-English view of the software side of this topic, see our software basics guide. Software is the layer users touch, while cloud computing is the delivery system that makes that software available from anywhere with a network connection.

Cloud Computing, What Is Cloud Computing? The Core Idea in Plain English

Cloud computing is the use of remote servers and software over the internet instead of local hardware. That is the simplest way to answer cloud computing, what is cloud computing, because the whole model comes down to renting computing resources from someone else’s data centers when you need them.

That setup matters because it separates ownership from usage. You can open an app, store a file, or run a database without buying the machine underneath it.

How Cloud Computing Works: Step by Step

Cloud computing works by separating where computing happens from where you use it. Your device sends a request over the internet, a cloud provider processes that request on its infrastructure, and the result comes back to you.

That infrastructure usually includes physical data centers filled with servers, storage devices, networking gear, and virtualization software. Virtualization lets one physical server act like many separate machines, which improves efficiency and lets a provider assign resources quickly.

The basic flow of a cloud request

Cloud computing follows a predictable chain of events, even though the exact setup changes by provider.

1. You open an app, upload a file, or make an application programming interface (API) request.
2. The request travels over the internet to the cloud provider.
3. The provider routes the request to the correct server or service.
4. The service processes the task and stores any output if needed.
5. The result returns to your device.

That flow is why cloud apps can sync across devices so easily. A photo edited on a phone can later appear on a laptop because both devices talk to the same remote system.

[IMAGE: Step-by-step cloud request flow from user device to cloud provider and back]

Virtualization, containers, and shared infrastructure

Cloud computing works well because providers divide physical hardware into smaller units that can be assigned on demand. Virtual machines, or VMs, are software-based computers that run on a physical server. Containers go one step further by packaging an app and its dependencies so it can run consistently across environments.

A practical example helps here. If a virtual machine is like renting a furnished apartment, a container is like renting a clean studio with only the essentials you need. Both are isolated, but containers are lighter and faster to start.

Why cloud computing feels flexible

Cloud computing feels flexible because resources can scale up or down when demand changes. If a website gets a traffic spike, the provider can add capacity. If demand drops, the system can release capacity.

That flexibility is one reason cloud spending has stayed high. Gartner estimated worldwide end-user spending on public cloud services at $679 billion in 2024, with continued growth expected in 2025 and beyond (Gartner, 2024).

Main Types of Cloud Computing

The main types of cloud computing are public cloud, private cloud, hybrid cloud, and multi-cloud. These models differ in who owns the infrastructure, who can access it, and how much control the customer keeps.

[IMAGE: Comparison graphic showing public, private, hybrid, and multi-cloud side by side]

Public cloud

Public cloud is cloud infrastructure owned and operated by a third-party provider and shared across many customers. You rent what you need, and the provider handles most of the hardware maintenance.

This model is usually the easiest way to start because it lowers upfront cost and shortens setup time. Public cloud is common for email, file storage, collaboration tools, web hosting, and AI services.

Private cloud

Private cloud is cloud infrastructure dedicated to one organization. It can live on-premises in the company’s own data center or in a hosted environment controlled for that one customer.

Private cloud gives more control over configuration and access, which matters for some financial, medical, or government workloads. The tradeoff is higher management effort and usually higher cost.

Hybrid cloud

Hybrid cloud combines public cloud and private cloud in one environment. A company might keep sensitive records in private infrastructure while using public cloud for web traffic, analytics, or backup.

This model works best when different workloads have different needs. Think of it like using a private safe for valuables and a public warehouse for bulk storage.

Multi-cloud

Multi-cloud means using services from more than one cloud provider. A business might use AWS for one app, Microsoft Azure for another, and Google Cloud for analytics.

Multi-cloud can reduce dependence on one provider, but it also adds complexity. Teams need more skills, more monitoring, and more careful cost control.

Cloud Computing Service Models

Cloud computing has three main service models: Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS). Each one gives you a different amount of control.

Infrastructure as a Service

Infrastructure as a Service gives you raw computing building blocks over the internet. You still configure the operating system and applications, but you do not buy the physical server.

This model is useful when you need flexibility and control without owning hardware. It is a common fit for developers, test environments, and websites with changing traffic.

Platform as a Service

Platform as a Service gives you a managed environment for building and deploying applications. The provider handles the underlying runtime, so your team can focus on code.

PaaS is helpful when you want to ship software faster without spending time on server maintenance. It is especially handy for small teams that want fewer operations tasks.

Software as a Service

Software as a Service gives you finished software through a browser or app. You log in, use the service, and the provider handles updates, security patches, and hosting.

SaaS is the cloud model most people already use every day. Email, video meetings, online documents, and music apps are all common examples.

Cloud Computing Benefits for Everyday Users and Businesses

Cloud computing lowers the cost and friction of using computing power. It lets people store files, run apps, and share data without maintaining the hardware themselves.

One major benefit is access from anywhere. If you have internet access, you can usually reach your files and tools from a phone, tablet, or laptop.

Another benefit is faster setup. A business can launch a website, app, or internal tool in hours or days rather than waiting for server purchases and installation.

Cloud computing also helps teams work together. Shared documents, synced storage, and hosted collaboration tools make it easier for people in different locations to work on the same material.

For content teams that publish online, cloud-based workflows often connect editing, analytics, and publishing tools. Our content analytics guide explains how teams measure what happens after content goes live.

Cost control

Cloud computing can reduce upfront spending because you do not need to buy every server before you use it. You move from capital spending on hardware to operating spending that tracks usage.

That does not automatically make the cloud cheap. Waste still happens when teams leave idle resources running, choose oversized instances, or store too much data in expensive tiers.

Practical ways to control cloud costs

• Set budgets, alerts, and spending limits before launching a workload.
• Tag resources by owner, project, department, and environment.
• Shut down unused test systems and remove unattached storage.
• Choose smaller resources first, then increase capacity when measurements justify it.
• Use automatic scaling for variable demand instead of running peak capacity all day.
• Move older data to a lower-cost storage tier when access is infrequent.
• Review data transfer paths, especially when services run across regions or providers.
• Compare the cost of managed services with the staff time required to operate them yourself.

For workloads that produce reports or process large datasets, connect cloud spending to business value. Our data analytics guide explains how measurement can help teams understand usage and outcomes.

Hidden and overlooked costs

The largest surprise is often not the virtual server itself. Data transfer out of the provider, premium support, managed database capacity, idle development environments, backup retention, monitoring, and software licenses can add material costs.

Pricing also varies by region, service tier, operating system, contract, and usage pattern. Provider pricing calculators are useful for estimates, but actual bills should be checked after deployment.

Common cloud billing patterns

• Pay as you go: You pay for the resources consumed during a billing period.
• Committed use: You agree to a term or spending commitment in exchange for a lower rate on eligible services.
• Reserved capacity: You reserve resources for a defined period, which can reduce cost when usage is predictable.
• Serverless or request-based pricing: You pay according to executions, requests, or processing time rather than keeping a server running continuously.
• Free tiers and trials: A provider offers limited usage at no charge, usually with quotas, eligibility rules, or expiration dates.

Free cloud services are useful for learning and small workloads, but “free” rarely means unlimited. Read the service limits and billing conditions before adding a payment method or moving production data.

Cloud Pricing, Billing, and Cost Optimization

Cloud pricing usually follows a usage-based model, but the bill may include more than the main service. Compute time, storage capacity, database use, data transfer, requests, licenses, support, and backup copies can all affect the final cost.

Scalability

Cloud computing scales more easily than local hardware because capacity can be added on demand. A startup can begin with a small setup and grow without rebuilding everything.

Scalability is one reason cloud adoption keeps expanding across industries. Flexera reported in its 2025 State of the Cloud survey that 84% of organizations have a multi-cloud strategy, which shows how common distributed cloud use has become (Flexera, 2025).

Reliability and backup

Cloud computing often improves reliability because providers build redundancy into their systems. Data can be copied across multiple servers or regions so one failure does not take everything down.

Backup and disaster recovery are easier too. Instead of keeping a single copy on one office machine, teams can store copies in separate locations and restore them when they need to.

Disaster recovery and business continuity

Disaster recovery is the process of restoring technology and data after a serious disruption. Business continuity is the broader plan for keeping essential operations running while systems are unavailable.

Define two practical targets:

• Recovery time objective (RTO): How long the business can operate without the system.
• Recovery point objective (RPO): How much recent data the business can afford to lose.

A small website may accept a longer recovery time, while an online retailer or emergency service may need automated failover. Choose the level of redundancy and backup spending based on the cost of downtime and data loss.

SLAs and realistic availability goals

A service-level agreement (SLA) is a provider’s written commitment about service availability or support. An SLA may offer service credits when a target is missed, but it does not guarantee that your application will remain available or compensate for every business loss.

Read the SLA’s definitions, exclusions, measurement method, claim process, and coverage limits. Your own uptime can be lower than the provider’s stated service level if your application has a faulty deployment, an expired certificate, a regional dependency, or a misconfigured network.

Redundancy and backups

Redundancy means keeping more than one working copy or component so one failure does not stop the service. Backups are separate copies used to restore data after deletion, corruption, ransomware, or another damaging event.

Cloud storage is not automatically a backup. If a file is deleted and synchronization removes it everywhere, the cloud copy may not help. A useful backup plan includes separate copies, retention periods, access controls, and regular restore tests.

For step-by-step recovery planning, see our backup and data recovery guide.

Regions and availability zones

A cloud region is a geographic area where a provider operates cloud infrastructure. An availability zone is a separate data-center location or infrastructure grouping within a region, designed to reduce the impact of certain failures.

Running an application in more than one availability zone can protect against a localized equipment or facility problem. Using more than one region can provide additional protection, but it may increase cost, complexity, latency, and data-sovereignty concerns.

Cloud Availability, Regions, Zones, SLAs, and Disaster Recovery

Cloud availability depends on how a workload is designed, not only on the provider’s reputation. Regions, availability zones, redundancy, backups, and tested recovery procedures all affect whether an application can keep operating after a failure.

Common Cloud Computing Risks and Tradeoffs

Cloud computing has real tradeoffs, and users should know them before moving important data or systems. The main concerns are cost, security, downtime, and vendor lock-in.

[IMAGE: A balance scale showing cloud benefits on one side and cloud tradeoffs on the other]

Cost creep

Cloud cost creep happens when usage grows faster than expected or when idle services keep running. A small bill can turn into a large one if no one watches storage, traffic, and compute usage.

The fix is simple in concept, though not always in practice: set budgets, tag resources, and shut down what you do not use. Cost calculators can help estimate spending before deployment, which is why tools like our AWS cost guide are useful for planning.

Security responsibility

Cloud providers secure the underlying infrastructure, but customers still protect their own accounts, data, and access rules. This division is often called the shared responsibility model.

That means strong passwords, multi-factor authentication, role-based access, and correct configuration still matter. A misconfigured storage bucket can expose data even if the provider’s infrastructure is secure.

Privacy and compliance

Privacy concerns how personal information is collected, used, shared, stored, and deleted. Compliance means meeting laws, regulations, contracts, or industry rules that apply to your data and organization.

Before choosing cloud services, check where data can be stored, which subprocessors may handle it, how long logs are retained, and how the provider supports deletion or export. A provider’s compliance certification can help, but it does not make your own applications or processes automatically compliant.

For sensitive workloads, review the provider’s security documentation, audit reports, encryption options, identity controls, breach-notification terms, and regional hosting choices. Legal and compliance requirements vary by country and industry, so a qualified adviser may be necessary for regulated data.

What the customer still needs to secure

Customers are usually responsible for choosing who can access data, configuring permissions, protecting login credentials, and deciding how information is stored and retained. A cloud provider cannot prevent every breach caused by a weak password or an accidentally public storage bucket.

• Use multifactor authentication for administrator and user accounts.
• Give each person only the access they need to do their work.
• Encrypt sensitive data while it moves and while it is stored.
• Keep applications, operating systems, and dependencies updated.
• Review access logs, unusual activity, and unused accounts.
• Test backups and recovery procedures instead of assuming they work.

For a broader explanation of threats, controls, and safer online practices, see our cybersecurity basics guide.

What cloud providers usually protect

Cloud providers generally secure the physical data centers, servers, networking equipment, and core services they operate. They also maintain parts of the platform, apply infrastructure patches, monitor facilities, and provide security tools.

The exact boundary depends on the service model. With SaaS, the provider manages most of the technology stack. With IaaS, you usually have more responsibility for the operating system, applications, configurations, and data.

Cloud Security, Privacy, Compliance, and Shared Responsibility

Cloud computing can be secure, but security is shared between the cloud provider and the customer. The provider protects the underlying cloud infrastructure, while you remain responsible for many decisions involving accounts, data, applications, and access.

Downtime and service limits

Cloud services can go down, just like any other online service. Most major providers publish status pages and service-level agreements, but no provider can promise zero outages.

Service limits also exist. A provider may cap requests, storage, or compute in certain situations, so teams need to design with those limits in mind.

Internet dependence and service outages

Most cloud services need a network connection. If your internet service fails, a provider has an outage, or a routing problem interrupts traffic, you may lose access even when your device is working normally.

Some apps offer offline access or local synchronization, but those features are not universal. If constant access matters, plan a fallback such as a second connection, local copies of essential files, or an alternative communication method.

Vendor lock-in

Vendor lock-in happens when a company becomes so dependent on one provider’s tools that switching is expensive or slow. Custom databases, proprietary APIs, and provider-specific automation can all make migration harder.

The best defense is to use portable formats when possible and document dependencies early. That does not eliminate lock-in, but it reduces the pain if plans change later.

Other practical limitations

Cloud services can also create configuration complexity, security mistakes, performance variation, and dependence on provider support. A service may be highly available overall while a particular region, account, or application configuration still fails.

Cloud computing is a tool, not a guarantee of lower cost, perfect uptime, or automatic security. The right choice depends on the workload, the people managing it, and the consequences of failure.

Data portability and migration difficulty

Data portability is the practical ability to export information in a usable format and move it elsewhere. Before committing to a service, confirm whether it supports standard export formats, bulk downloads, transfer tools, and reasonable exit terms.

• Identify which data must remain portable.
• Ask how quickly a complete export can be produced.
• Check whether exported data includes metadata, permissions, and version history.
• Estimate transfer charges, storage costs, and staff time during a move.
• Test a small export before the data becomes difficult to handle.

Vendor lock-in

Vendor lock-in happens when moving an application or dataset to another provider becomes difficult or expensive. Lock-in can result from proprietary databases, provider-specific APIs, custom identity systems, unusual data formats, or applications designed around one platform.

Lock-in is not always bad. A managed service can save development time and improve productivity. The important point is to choose deliberately rather than discovering later that a migration would require a full rewrite.

Cloud Risks and Disadvantages

Cloud computing reduces the need to own hardware, but it does not remove technology risk. The main disadvantages include service outages, internet dependence, unexpected bills, provider dependence, and the work required to move data later.

Cloud Computing in AI and Machine Learning

Cloud computing is a common base layer for artificial intelligence (AI) and machine learning because training and serving models need lots of compute power. Most general-purpose laptops are not built for that workload.

AI and machine learning systems often rely on graphics processing units (GPUs), which are specialized chips that handle parallel computation efficiently. Cloud providers let teams rent GPU time instead of buying expensive hardware outright.

This matters even for non-technical users because many apps now use cloud-based AI behind the scenes. Image generation, transcription, recommendation systems, and chat tools often run on cloud infrastructure.

[IMAGE: AI model workflow showing data upload, training on cloud GPUs, and deployment to an app]

Cloud computing also supports MLOps, which means machine learning operations. MLOps is the process of building, testing, deploying, and monitoring machine learning models in a repeatable way.

If you want to understand one of the model-tuning concepts used in that workflow, our hyperparameter tuning guide explains how teams adjust model settings to improve results.

Real-World Cloud Computing Examples

Cloud computing appears in daily life more often than people notice. You probably use it every day when you check email, back up photos, watch video, or work in online documents.

Email is one of the clearest examples. Instead of storing your mailbox on your personal device, the service keeps it on remote servers so you can access it from multiple devices.

Streaming is another simple example. When you watch a movie on a phone or TV, the media file comes from cloud infrastructure that can deliver it fast to millions of users.

Businesses use cloud computing for customer databases, payroll, analytics, app hosting, and customer support systems. Many smaller companies now run nearly all office software through the cloud because it reduces maintenance overhead.

How to Choose the Right Cloud Setup

Choosing the right cloud setup starts with your goal. If you want email and file sharing, SaaS is usually enough. If you want to build software, PaaS or IaaS may fit better.

Your choice should also reflect your data sensitivity, team skills, budget, and expected traffic. A school, startup, hospital, and bank may all use cloud computing, but they will not choose the same structure.

A simple decision guide

Use these questions to narrow the choice.

1. Do you want ready-made software? Choose SaaS.
2. Do you want to build and deploy apps with less server work? Choose PaaS.
3. Do you need more control over systems and configuration? Choose IaaS.
4. Do you need strict separation for sensitive data? Consider private or hybrid cloud.
5. Do you expect multiple tools from different providers? Consider multi-cloud.

Common migration strategies

Migration strategies describe how much an application changes during the move. The best option can differ between workloads in the same organization.

Assess each workload first

A workload is an application, service, database, or related group of systems that performs a business task. For every workload, document its purpose, users, data, dependencies, performance needs, security requirements, and recovery targets.

• What business process depends on it?
• Which databases, files, APIs, or identity systems does it use?
• How much downtime is acceptable during migration?
• Does the data have location, retention, or compliance requirements?
• Will usage be steady, seasonal, or unpredictable?
• What will the cloud design cost at normal and peak usage?
• How will the team monitor, secure, update, and recover it afterward?

Cloud Migration Basics: Planning and Workload Assessment

Cloud migration is the process of moving applications, data, and supporting systems from local infrastructure or one provider to another. Successful migration starts with assessment and testing, not with moving everything at once.

Choosing by audience and scenario

For individuals: SaaS is usually the simplest choice for email, documents, photo synchronization, and subscriptions. Check storage limits, privacy settings, export options, and whether offline access is available.

For small businesses: Managed SaaS or PaaS can reduce the need for dedicated IT staff. A public cloud is often practical for a website or online application, provided the business sets budgets, protects administrator accounts, and maintains independent backups.

For enterprises: Hybrid or multi-cloud may be appropriate when regulations, existing systems, resilience requirements, or regional operations influence the design. These models require stronger governance, monitoring, identity management, and cost controls.

• File storage: Start with SaaS or managed cloud storage when easy sharing and synchronization matter.
• Website hosting: Use managed hosting or PaaS for simplicity; choose IaaS when you need operating-system-level control.
• App development: Choose PaaS or containers when developers want to focus on code rather than server maintenance.
• Backup: Select a service with clear retention, restore testing, encryption, and export features.
• Analytics: Compare data warehouse, storage, processing, and data-transfer costs before selecting a platform.
• AI workloads: Check accelerator availability, data privacy, model services, regional restrictions, and usage-based pricing.

Comparing major cloud providers

AWS, Microsoft Azure, and Google Cloud all provide infrastructure, storage, databases, analytics, security tools, and developer services. Their product names, pricing structures, regional offerings, and management tools differ, so a provider should be evaluated against a specific workload rather than chosen by reputation alone.

This comparison is based on provider documentation and product information available in 2026; services, prices, regions, and terms change over time. Check each provider’s current pricing pages, region lists, service documentation, SLA terms, and support plans before making a purchase.

Deployment model or service model?

These terms describe different decisions. A deployment model describes where infrastructure is deployed and who shares it. Public, private, hybrid, and multi-cloud are deployment models.

A service model describes how much the provider manages for you. IaaS gives you more control, PaaS manages more of the application platform, and SaaS provides finished software.

How to Choose a Cloud Provider and Service Model

Choose a cloud provider by matching its services, location, pricing, support, skills, and security controls to your workload. Choose a service model by deciding how much infrastructure management your team wants to handle.

Common Mistakes to Avoid with Cloud Computing

Cloud computing works best when people plan for cost, security, and usage patterns before they move anything important. The biggest mistakes are usually simple, not technical.

Leaving resources on

The mistake is keeping test servers, unused storage, or idle databases running. That is a billing problem because cloud pricing is often usage-based, so unused resources still cost money.

Turn off what you do not need, automate shutdown schedules, and review billing regularly.

Treating security as the provider’s job

The mistake is assuming the cloud provider handles everything. That is a security problem because the provider secures the platform, but customers still manage identities, passwords, permissions, and data access.

Use multi-factor authentication, limit admin access, and check configurations after every major change.

Ignoring data backups

The mistake is storing important files in one place and assuming they are safe. That is risky because accidental deletion, ransomware, or account problems can still cause data loss.

Keep backups in a separate location and test recovery before you need it.

Choosing the wrong service model

The mistake is buying too much control or too little control. That leads to wasted time because IaaS, PaaS, and SaaS solve different problems.

Match the service model to the job rather than choosing the one with the most features.

Frequently Asked Questions About Cloud Computing

What is cloud computing in simple terms?

Cloud computing is using storage, software, and computing power over the internet instead of keeping everything on your own device. It lets you access services from anywhere without owning the hardware behind them.

Is cloud computing the same as online storage?

No, online storage is only one part of cloud computing. Cloud computing also includes servers, databases, apps, analytics, and AI tools that run remotely.

What are the three main types of cloud computing?

The three main types are public cloud, private cloud, and hybrid cloud. Public cloud shares infrastructure across many customers, private cloud is dedicated to one organization, and hybrid cloud combines both.

Why do companies use cloud computing?

Companies use cloud computing because it can lower upfront hardware spending, speed up deployment, and make scaling easier. It also helps teams collaborate across locations and support remote work.

Is cloud computing safe?

Cloud computing can be safe when accounts are configured well and access is controlled. The biggest risks usually come from weak passwords, poor permissions, and exposed data, not from the cloud model itself.

Can cloud computing work without the internet?

Most cloud computing depends on internet access because the resources live on remote servers. Some apps can keep limited offline features, but they usually sync with the cloud once the connection returns.

Who should learn cloud computing first?

Anyone who uses modern software, manages a website, or works with data should learn the basics. Beginners do not need to become engineers, but they should understand the main models, costs, and risks.

Key Takeaways

• Cloud computing is internet-delivered access to storage, software, and computing power.
• The three main service models are IaaS, PaaS, and SaaS, and each one gives you a different amount of control.
• Public, private, hybrid, and multi-cloud setups solve different cost, security, and flexibility needs.
• Cloud computing is a major foundation for AI and machine learning because it provides scalable compute, storage, and GPU access.
• Smart cloud use depends on cost tracking, access control, backups, and choosing the right service model for the job.

More on this topic

A safer migration sequence

Begin with an inventory and a small pilot. Establish identity controls, logging, backup, monitoring, and cost alerts before moving important production data.

1. Set business goals and define success measures.
2. Map applications, data, dependencies, and owners.
3. Classify data by sensitivity, retention, and location requirements.
4. Choose a migration strategy for each workload.
5. Estimate costs and design the target cloud environment.
6. Test performance, security, backups, and restoration.
7. Move a low-risk workload first and record what went wrong.
8. Schedule production migration with a rollback plan.
9. Monitor the new environment and remove old infrastructure only after validation.

Do not treat migration as the finish line. After the move, review permissions, resource sizes, service reliability, application performance, and monthly bills. A well-planned cloud migration should improve the way a workload operates, not merely place the same problems on a different server.