Andina tent wants to move its application to the cloud. The first decision is where to deploy and what part of the service should continue to work with a malfunction.
Key concepts
The cloud allows to supply resources by demand with console, API or tools. Scalability is the capacity to increase or reduce resources elasticity It's tipping them as demand. A high-dimensional and always light VM can be scalable without using elasticity. To pay for use also doesn't mean to pay only when there are customers visiting your application: the Meter can register supplied capacity.
In IaaS you manage, among other things, your VMs' operating system and the application. Section 3 PaaS, Oracle manages more components of the platform. In SaaS, you're using an end-up application. The more the service is run, the more tasks are distributed. The responsibility for data and access remains vital.
OCI gathers compute, network, storage, databases, security and other services. One tenancy That's the environment of your organisation. A OCID uniquely identifies a resource; its display name helps you recognize it but does not replace that identifier.
Failed regions and domains
One region It's a geographical area. Contains one or more Availability Domains (AD)Each of them includes one or more data centres. One AD contains three Fault Domains (FD): Hardware combinations that allow to separate instances from malfunctions and maintenance. Not all regions have 3 AD. Query regional structure of OCI.
Two different FD VMs within the same AD help with some hardware malfunctions. To support AD loss you have to design another AD available. for a regional event, in another region. In both cases you also have to solve data, connectivity and recovery. Double only VM does not guarantee continuity.
Choose region by user closeness, residence requirements, available services and capacity. The high availability seeks to maintain service; and disaster recovery defines how to re-establish it after a major interruption. Think about both the tolerable recovery time and how many data you could lose.
Console and how to consume OCI
Before to look for a VM on the console, verifies tenancy, region and compartments. Many resources are regional or associated with an AD. The compartments are logical groupings that can contain resources from different regions: changing compartments does not change the console region.
On the console tour you identify the service menu, search, region selector, compartment selector, identity profile, help and billing tools. The console, CLI, SDK and API have access to the same services as IAM. A resource doesn't change his own because you created him with a different interface.
The Public cloud It houses resources in regions operated by Oracle. OCI Dedicated Region Take a OCI region to the client's data centre under service conditions. Roving Edge Infrastructure about compute and storage to remote locations. On a stage multicloud, solutions such as Oracle Interconnect for Azure and Oracle Database @ Azure connect or integrate environments from different providers. They are different needs: location, border operation and intercloud integration. Query la distributed cloud supply.
Step-by-step example
The store wants to keep the web available if a server fails. It has yet to demand continuity as a result of the loss of a region.
- The team compares regions close to its customers and checks the availability of necessary services.
- It creates a compartments Productionand different from that of learning.
- It distributes two instances between FD different from an AD and puts a balancing machine in front.
- Separates persistent life cycle data from a single instance and sets back up copies.
- It simulates conceptual loss of an instance: the balancing machine should send new requests to a healthy backend. A loss of AD requires an additional design.
To guide you at a demonstration, write down OCID, region, AD as appropriate, state and resource. If a search does not return results, check that context and permissions before creating a duplicate.
Common errors
- "Two instances mean high availability." If they depend on a single fault point, they can both become unusable. It includes balancing, data and health of the application.
- "A compartments is a data centre." Organizes and delimits permissions doesn't provide physical separation by itself.
- "All regions have 3 AD." The quantity's different. The three FDs belong to every AD.
Retirement exercise
An application should continue to be available if a hardware group suffers maintenance. The region has a single AD. A partner propose to create a second compartments and leave the two VMs with the same FD. What would you change and what risk would she stay?
Solution
He would distribute the instances between different FD and check the balancing and data availability. The second compartments can serve as an organization or permissions but does not change its physical placement. AD remains a shared fault area. A more comprehensive continuity requirement could require another region and a replication and recovery strategy.
Sources and review
Reg regions, AD and FD, the OCI console y distributed cloud. Without looking at text: why does user distance matter? What information do you need to find a VM? When did Dedicated Region and Multicloud meet different requirements?