This is a continuation of an article that describes operational considerations for hosting solutions on Azure public cloud.   

There are several references to best practices throughout the series of articles we wrote from the documentation for the Azure Public Cloud. The previous article focused on the antipatterns to avoid, specifically the busy frontend antipattern. This one focuses on monolithic persistence antipattern.

This antipattern occurs when a single data store hurts performance due to resource contention. Additionally, the use of multiple data sources can help with virtualization of data and query.

A specific example of this antipattern is when applications save transactional records, logs, metrics and events to the same database. The online transaction processing benefits from a relational store but logs and metrics can be moved to a log index store and time-series database respectively. Usually, a single datastore works well for transactional data but this does not mean documents need to be stored in the same data store. An object storage or document database can be used in addition to a regular transactional database to allow individual documents to be shared without any impact to the business operations. Each document can then have its own web accessible address.

This antipattern can be fixed in one of several ways. First, the data types must be listed, and their corresponding data stores must be assigned. Many data types can be bound to the same database but when they are different, they must be passed to the data stores that handles them best. Second, the data access patterns for each data type must be analyzed.  If the data type is a document, a CosmosDB instance is a good choice. Third, if the database instance is not suitable for all the data access patterns of the given data type, it must be scaled up. A premium sku will likely benefit this case.

Detection of this antipattern is easier with the monitoring tools and the built-in supportability features of the database layer. If the database activity reveals significant processing, contention and very low data rate, it is likely that this antipattern is manifesting.

Examine the work performed by the database in terms of data types which can be narrowed down by callers and scenarios, may reveal just the culprits that are likely to be causing this antipattern

Finally, periodic assessments must be performed on the data storage tier.

 

 

This is a continuation of an article that describes operational considerations for hosting solutions on Azure public cloud.   

There are several references to best practices throughout the series of articles we wrote from the documentation for the Azure Public Cloud. The previous article focused on the antipatterns to avoid, specifically the busy database antipattern. This one focuses on busy frontend antipattern.

This antipattern occurs when there are many background threads that can starve foreground tasks of their resources which decreases response times to unacceptable levels. There is a lot of advantages to running background jobs which avoids the interactivity for processing and can be scheduled asynchronously. But the overuse of this feature can hurt performance due to the tasks consuming resources that foreground workers need for interactivity with the user, leading to a spinning wait and frustrations for the user. It appears notably when the foreground is monolithic compressing the business tier with the application frontend. Runtime costs might shoot up if this tier is metered. An application tier may have finite capacity to scale up. Compute resources are better suitable for scale out rather than scale up and one of the primary advantages of a clean separation of layers and components is that they can be hosted even independently. Container orchestration frameworks facilitate this very well. The Frontend can be as lightweight as possible and built on model-view-controller or other such paradigms so that they are not only fast but also hosted on separate containers that can scale out.

This antipattern can be fixed in one of several ways. First the processing can be moved out of the application tier into an Azure Function or some background api layer. If the application frontend is confined to data input and output display operations using only the capabilities that the frontend is optimized for, then it will not manifest this antipattern. APIs and Queries can articulate the business layer interactions. The application then uses the .NET framework APIs to run standard query operators on the data for display purposes.

UI interface is designed for purposes specific to the application. The introduction of long running queries and stored procedures often goes against the benefits of a responsive application. If the processing is already under the control of the application techniques, then they should not be moved. 

Avoiding unnecessary data transfer solves both this antipattern as well as chatty I/O antipattern. When the processing is moved to the business tier, it provides the opportunity to scale out rather than require the frontend to scale up.

Detection of this antipattern is easier with the monitoring tools and the built-in supportability features of the application layer. If the frontend activity reveals significant processing and very low data emission, it is likely that this antipattern is manifesting.

Examine the work performed by the Frontend in terms of latency and page load times which can be narrowed down by callers and scenarios, may reveal just the view models that are likely to be causing this antipattern

Finally, periodic assessments must be performed on the application tier.

 

This is a continuation of an article that describes operational considerations for hosting solutions on Azure public cloud.

There are several references to best practices throughout the series of articles we wrote from the documentation for the Azure Public Cloud. The previous article focused on the antipatterns to avoid, specifically the cloud readiness antipatterns. This article focuses on the extraneous fetching antipattern.

When services call datastores, they retrieve data for a business operation, but they often result in unnecessary I/O overhead and reduced responsiveness.  This antipattern can occur if the application is trying to save on the number of requests by fetching more than required. This is a form of overcompensation and is commonly seen with catalog operations because the filtering is delegated to the middle tier.  For example, user may need to see a subset of the details and probably does not need to see all the products at once yet a large dataset from the catalog is retrieved.  Even if the user is browsing the entire catalog, paginating the results avoids this antipattern.

Another example of this problem is the inappropriate choices in design or code where for example, a service gets all the product details via the entity framework and then filters only a subset of the fields while discarding the rest. Yet another example is when the application retrieves data to perform an aggregation that could be done by the database instead. The application calculates total sales by getting every record for all orders sold instead of executing a query where the predicates are pushed down to the store. Similarly other manifestations might come about when the EntityFramework uses LINQ to entities. In this case, the filtering is done in memory by retrieving the results from the table because a certain method in the predicate could not be translated to a query. The call to AsEnumerable is a hint that there is a problem because the filtering based on IEnumerable is usually done on the client side rather than the database. The default for LINQ to Entities is IQueryable which pushes the filters to the data source.

Fetching only the relevant columns from a table as compared to fetching all the columns is another classic example of this antipatterns and even though this might have worked when the table was only a few columns wide, it changes the game when the table adds several more columns. Similarly, aggregation performed in the database overcomes this antipattern instead of doing it in memory on the application side.

As with data access best practice, some considerations for performance holds true here as well. Partitioning data horizontally may reduce contention. Operations that support unbounded queries can implement pagination. Features that are built right into the data store can be leveraged. Some calculations need not be repeated especially with summation forms. Queries that return a lot of results can be further filtered. Not all operations can be offloaded to the database but those where the database is highly optimized can be offloaded.

A few ways to detect this antipattern include identifying slow workloads or transactions, behavioral patterns exhibited by the system due to limits, correlating the instances of slow workloads with those patterns, identifying the data stores being used, identify any slow running queries that reference these data source and performing a resource specific analysis of how the data is used and consumed.

These are some of the ways to mitigate this antipattern.

Some of the metrics that help with detecting and mitigation of extraneous fetching antipattern include total bytes per minute, average bytes per transaction and requests per minute.

 

This is a continuation of an article that describes operational considerations for hosting solutions on Azure public cloud. 

There are several references to best practices throughout the series of articles we wrote from the documentation for the Azure Public Cloud. The previous article focused on the antipatterns to avoid, specifically the Chatty I/O antipattern. This one focuses on improper instantiation antipattern

When new instances of classes are continually created instead of once, they can have a significant impact on performance and responsiveness. Connections and clients are significantly costly resources to setup. They must be created once and reused. Each connection or client instantiation requires server handshakes which not only incur network delay but also involve memory usage, and the cumulative effect of numerous setup requests can slow down the system. There are some common causes of improper instantiation which include:

Connections and clients are created for the purpose of a data access request. When they are scoped to one request for the sake of cleanup, they involve the server to respond. Reading and writing individual records to a database as distinct requests – When records are often fetched one at a time, then a series of queries are run one after the other to get the information. It is exacerbated when the shared libraries use hides this behavior and each access request recreates a connection or a client. The same might happen on write requests.

Implementing a single logical operation as a series of data access requests. This occurs when objects use wrappers for connections and clients and they are scoped to methods invoking them which results in connections and clients to be disposed often. The code appears as if a wrapper is used locally when in fact every instantiation of the wrapper is coming with at least the cost of the RTT. When there are many networks round trips, the cost is cumulative and even prohibitive. It is easily observable when a wrapper has many instantiations, and each time it creates a connection or client. In such case, there is also the requirement to perform validation after every access.

Reading and writing to a file on disk – File I/O also hides the distributed nature of interconnected file systems.  Every byte written to a file on a mount must be relayed to the original on the remote server. When the writes are several, the cost accumulates quickly. It is even more noticeable when the writes are only a few bytes and frequent. If each access requires its own connection or client, the application might not even know the high number of connections it is making.

There are several ways to fix the problem. They are about detection and remedy. When the number of server handshakes are many, they can be batched into reused connections via shared clients or connection pooling. The database can be read with a shared and reusable connection pool rather than a single connection. It also provides an opportunity for the database to free up memory corresponding to client connections. Web APIs can be designed with the REST best practices. Instead of separate GET method for different properties there can be single GET method for the resource representing the object.

When more information is retrieved via fewer connection and client instantiations, there is a risk of falling into the extraneous fetching antipattern by trying to prefetch more than is necessary. The right tradeoff depends on the usages. It is also important to read only as much as necessary to avoid both the size and the frequency of connections. Sometimes, connections and clients can also be involving a mixed mode, shared for accounts with most requests and dedicated for everything else. When connection is reused from a shared pool, they need not be locked at too large a scope or for longer duration. 

 

 

 

 

 

This is a continuation of an article that describes operational considerations for hosting solutions on Azure public cloud.

There are several references to best practices throughout the series of articles we wrote from the documentation for the Azure Public Cloud. The previous article focused on the antipatterns to avoid, specifically the extraneous fetching antipattern. This one focuses on the Chatty I/O antipattern.

When I/O requests are frequent and numerous, they can have a significant impact on performance and responsiveness. Network calls and other I/O operations are much slower compared to compute tasks. Each I/O request has a significant overhead as it travels up and down the networking stack on local and remote and includes the round trip time, and the cumulative effect of numerous I/O operations can slow down the system. There are some common causes of chatty I/O which include:

Reading and writing individual records to a database as distinct requests – When records are often fetched one at a time, then a series of queries are run one after the other to get the information. It is exacerbated when the Object-Relational Mapping hides the behavior underneath the business logic and each entity is retrieved over several queries. The same might happen on write for an entity.

Implementing a single logical operation as a series of HTTP requests. This occurs when objects residing on a remote server are represented as proxy in the memory of the local system. The code appears as if an object is modified locally when in fact every modification is coming with at least the cost of the RTT. When there are many networks round trips, the cost is cumulative and even prohibitive. It is easily observable when a proxy object has many properties, and each property get / set requires a relay to the remote object. In such case, there is also the requirement to perform validation after every access.

Reading and writing to a file on disk – File I/O also hides the distributed nature of interconnected file systems.  Every byte written to a file on a mount must be relayed to the original on the remote server. When the writes are several, the cost accumulates quickly. It is even more noticeable when the writes are only a few bytes and frequent.

There are several ways to fix the problem. They are about detection and remedy. When the number of I/O requests are many, they can be batched into coarse requests. The database can be read with one query substituting many queries. It also provides an opportunity for the database to execute it better and faster. Web APIs can be designed with the REST best practices. Instead of separate GET method for different properties there can be single GET method for the resource representing the object. Even if the response body is large, it will likely be a single request. File I/O can be improved with buffering and using cache. Files may need not be opened or closed repeatedly. This helps to reduce fragmentation of the file on disk.

When more information is retrieved via fewer I/O calls, there is a risk of falling into the extraneous fetching antipattern. The right tradeoff depends on the usages. It is also important to read only as much as necessary to avoid both the size and the frequency of calls. Sometimes, data can also be partitioned into two chunks, frequently accessed data that accounts for most requests and less frequently accessed data that is used rarely. When data is written, resources need not be locked at too large a scope or for longer duration.

 

 

 

This is a continuation of an article that describes operational considerations for hosting solutions on Azure public cloud.

There are several references to best practices throughout the series of articles we wrote from the documentation for the Azure Public Cloud. The previous article focused on the antipatterns to avoid, specifically the cloud readiness antipatterns. This one talks about design principles and advanced operations.

A management baseline provides a minimum level of business commitment for all supported workloads. It includes a standard business commitment to minimize business interruptions and accelerate recovery if service is interrupted. Usually it includes inventory and visibility, operational compliance, and protection and recovery – all of which provide streamlined operational management. It does not apply to mission critical workloads, but it covers 80% of the less critical workloads.

There are a few ways to go beyond the management baseline which includes enhanced baseline, platform specialization, and workload specialization.

The enhanced management baseline uses cloud-native tools to improve uptime and decrease recovery times. It significantly reduces cost and implementation time.

The management specialization are aspects of workload and platform operations which require changes to design and architecture principles, and these could take time and result in increased operating expenses. The enhanced management baseline applies broadly to many workloads while this one applies specifically to certain cases. There are two areas of specialization: 1) the platform specialization and 2) workload specializations. The former resolves key pain points in the platform and distributes the investments across multiple workloads and the latter involves ongoing operations of a specific mission-critical workload.

In addition to these management baselines, there are a few steps that apply to each specialization process. These include improved system design, automated remediation, scaled solution, and continuous improvement. Improved system design is the most effective approach among these, and it applies universally to most operations of any platform. It increases stability and decreases impact from changes in business operations. Both the Cloud Adoption Framework and the Azure Well-architected framework provide guiding tenets for improving the quality of a platform or a specific workload with the five pillars of architecture excellence which include cost optimization, operational excellence, performance efficiency, reliability, and security.

Business interruptions cause technical debt and if it cannot be automatically resolved, automated remediation is an alternative. Use of Azure automation and Azure Monitor can detect trends and provide automated remediation which is the most common approach. Similarly, a service catalog can list applications that can be deployed for internal consumption. A platform can then maximize adoption and minimize maintenance overhead with the use of the service catalog.

 

 

 

This is a continuation of an article that describes operational considerations for hosting solutions on Azure public cloud.

There are several references to best practices throughout the series of articles we wrote from the documentation for the Azure Public Cloud. The previous article focused on the antipatterns to avoid, specifically the cloud readiness antipatterns. This article focuses on the no-caching antipattern.

 A no-caching antipattern occurs when a cloud application handles many concurrent requests, and they fetch the same data. Since there is contention for the data access, it can reduce performance and scalability. When the data is not cached, it leads to many manifestations of areas for improvement.

First, the fetching of data can traverse several layers and go deep into the stack taking significant resource consumption and increasing costs in terms of I/O overhead and latency. It repeatedly constructs the same objects or data structures.

Second, it makes excessive calls to a remote service that has a service quota and throttles clients past a certain limit.

Both these can lead to degradation in response times, increased contention, and poor scalability.

The examples of no-caching antipattern are easy to spot. Entity framework calls that are repeatedly called for the same read-only data fits this antipattern. The use of a cache might have simply been overlooked but usually the case is that the cache could not be included in the design because of some unknowns. The benefits and drawbacks of using a cache is not clear then. There might be a concern about the accuracy and the freshness of the cached data.

Other times, the cache was left out because the application was migrated from on-premises where network latency and response times were controlled. The system might have been running on expensive high-performance hardware unlike the commodity cloud virtual machine scale sets.

Rarely, it might even be the case where the caching was simply left out of the architecture design and for operations to include via standalone independent products which was not clearly communicated. Other times, the introduction of a cache might increase latency, maintenance and ownership and decrease overall availability. It might also interfere with existing caching strategies and expiration policies of the underlying systems. Some might prefer to not add an external cache to a database and only as a sidecar for the web services. It’s true that databases can cache even materialized views for a connection, but the addition of a cache lookup could be cheap in all cases where the compute in the deeper systems could be costly and can be avoided.

There are two strategies to fix the problem. The first one includes the on-demand network or cache-aside strategy. When the application tries to read the data from the cache, and if it isn’t there, it retrieves and puts it in the cache. When the application writes the change directly to the data source, it removes the old value from the source but refilled the next time it is required.

Another strategy might be to always keep static resources in the cache with no expiration date. This is equivalent to CDN usage although CDNs are for distribution.  Applications that cache dynamic data should be designed to support eventual consistency.

No matter how the cache is implemented, it must support fallback to the deep data access when the data is not available in the cache. This Circuit-breaker pattern merely avoids overwhelming the data source.