I want to extract multiple entities from a user input.
Example- "Service httpd is not responding because of high CPU usage and DNS Error"
So here I want to identify below:
Httpd
High CPU usage
DNS Error
And I will be using this keywords to get a response from a Database.
Just annotate them accordingly, e.g.
## intent: query_error
- Service [httpd](keyword) is not responding because of [high CPU usage](keyword) and [DNS Error](keyword)
Having the sentence from above, Rasa NLU would extract 3 entities of type keyword. You can then access these entities in a custom action and query your database.
Regarding the number of examples which are required: this depends on
the NLU pipeline which you are using. Typically tensorflow_embedding requires more training examples than spacy_sklearn since it does not use pretrained language models.
the number of different values your entities can have. If it is only httpd, high CPU usage, and DNS error then you don't need a lot of examples. However, if you have a thousand different values for your entity, then you need more training examples
One intent is enough if you always want to trigger the same custom action. However, if you want to classify different type of problems, e.g. server problems and client problems, and trigger different databases depending on the type of problems, you might consider having multiple intents.
Sorry for the vague answers, but in machine learning most things are highly dependent on the use case and the dataset.
Related
I am investigating options to build a system to provide "Entity Access Control" across a microservices based architecture to restrict access to certain data based on the requesting user. A full Role Based Access Control (RBAC) system has already been implemented to restrict certain actions (based on API endpoints), however nothing has been implemented to restrict those actions against one data entity over another. Hence a desire for an Attribute Based Access Control (ABAC) system.
Given the requirements of the system to be fit-for-purpose and my own priorities to follow best practices for implementations of security logic to remain in a single location I devised to creation of an externalised "Entity Access Control" API.
The end result of my design was something similar to the following image I have seen floating around (I think from axiomatics.com)
The problem is that the whole thing falls over the moment you start talking about an API that responds with a list of results.
Eg. A /api/customers endpoint on a Customers API that takes in parameters such as a query filter, sort, order, and limit/offset values to facilitate pagination, and returns a list of customers to a front end. How do you then also provide ABAC on each of these entities in a microservices landscape?
Terrible solutions to the above problem tested so far:
Get the first page of results, send all of those to the EAC API, get the responses, drop the ones that are rejected from the response, get more customers from the DB, check those... and repeat until either you get a page of results or run out of customers in the DB. Tested that for 14,000 records (which is absolutely within reason in my situation) would take 30 seconds to get an API response for someone who had zero permission to view any customers.
On every request to the all customers endpoint, a request would be sent to the EAC API for every customer available to the original requesting user. Tested that for 14,000 records the response payload would be over half a megabyte for someone who had permission to view all customers. I could split it into multiple requests, but then you are just balancing payload size with request spam and the performance penalty doesn't go anywhere.
Give up on the ability to view multiple records in a list. This totally breaks the APIs use for customer needs.
Store all the data and logic required to perform the ABAC controls in each API. This is fraught with danger and basically guaranteed to fail in a way that is beyond my risk appetite considering the domain I am working within.
Note: I tested with 14,000 records just because its a benchmark of our current state of data. It is entirely feasible that a single API could serve 100,000 or 1m records, so anything that involves iterating over the whole data set or transferring the whole data set over the wire is entirely unsustainable.
So, here lies the question... How do you implement an externalised ABAC system in a microservices architecture (as per the diagram) whilst also being able to service requests that respond with multiple entities with a query filter, sort, order, and limit/offset values to facilitate pagination.
After dozens of hours of research, it was decided that this is an entirely unsolvable problem and is simply a side effect of microservices (and more importantly, segregated entity storage).
If you want the benefits of a maintainable (as in single piece of externalised infrastructure) entity level attribute access control system, a monolithic approach to entity storage is required. You cannot simultaneously reap the benefits of microservices.
CQRS states: command should not query read side.
Ok. Let's take following example:
The user needs to create orders with order lines, each order line contains product_id, price, quantity.
It sends requests to the server with order information and the list of order lines.
The server (command handler) should not trust the client and needs to validate if provided products (product_ids) exist (otherwise, there will be a lot of garbage).
Since command handler is not allowed to query read side, it should somehow validate this information on the write side.
What we have on the write side: Repositories. In terms of DDD, repositories operate only with Aggregate Roots, the repository can only GET BY ID, and SAVE.
In this case, the only option is to load all product aggregates, one by one (repository has only GET BY ID method).
Note: Event sourcing is used as a persistence, so it would be problematic and not efficient to load multiple aggregates at once to avoid multiple requests to the repository).
What is the best solution for this case?
P.S.: One solution is to redesign UI (more like task based UI), e.g.: User first creates order (with general info), then adds products one by one (each addition separate http request), but still I need to support bulk operations (api for third party applications as an example).
The short answer: pass a domain service (see Evans, chapter 5) to the aggregate along with the other command arguments.
CQRS states: command should not query read side.
That's not an absolute -- there are trade offs involved when you include a query in your command handler; that doesn't mean that you cannot do it.
In domain-driven-design, we have the concept of a domain service, which is a stateless mechanism by which the aggregate can learn information from data outside of its own consistency boundary.
So you can define a service that validates whether or not a product exists, and pass that service to the aggregate as an argument when you add the item. The work of computing whether the product exists would be abstracted behind the service interface.
But what you need to keep in mind is this: products, presumably, are defined outside of the order aggregate. That means that they can be changing concurrently with your check to verify the product_id. From the point of view of correctness, there's no real difference between checking the validity of the product_id in the aggregate, or in the application's command handler, or in the client code. In all three places, the product state that you are validating against can be stale.
Udi Dahan shared an interest observation years ago
A microsecond difference in timing shouldn’t make a difference to core business behaviors.
If the client has validated the data one hundred milliseconds ago when composing the command, and the data was valid them, what should the behavior of the aggregate be?
Think about a command to add a product that is composed concurrently with an order of that same product - should the correctness of the system, from a business perspective, depend on the order that those two commands happen to arrive?
Another thing to keep in mind is that, by introducing this check into your aggregate, you are coupling the ability to change the aggregate to the availability of the domain service. What is supposed to happen if the domain service can't reach the data it needs (because the read model is down, or whatever). Does it block? throw an exception? make a guess? Does this choice ripple back into the design of the aggregate, and so on.
I am currently building a microservices-based application developed with the mean stack and am running into several situations where I need to share models between bounded contexts.
As an example, I have a User service that handles the registration process as well as login(generate jwt), logout, etc. I also have an File service which handles the uploading of profile pics and other images the user happens to upload. Additionally, I have an Friends service that keeps track of the associations between members.
Currently, I am adding the guid of the user from the user table used by the User service as well as the first, middle and last name fields to the File table and the Friend table. This way I can query for these fields whenever I need them in the other services(Friend and File) without needing to make any rest calls to get the information every time it is queried.
Here is the caveat:
The downside seems to be that I have to, I chose seneca with rabbitmq, notify the File and Friend tables whenever a user updates their information from the User table.
1) Should I be worried about the services getting too chatty?
2) Could this lead to any performance issues, if alot of updates take place over an hour, let's say?
3) in trying to isolate boundaries, I just am not seeing another way of pulling this off. What is the recommended approach to solving this issue and am I on the right track?
It's a trade off. I would personally not store the user details alongside the user identifier in the dependent services. But neither would I query the users service to get this information. What you probably need is some kind of read-model for the system as a whole, which can store this data in a way which is optimized for your particular needs (reporting, displaying together on a webpage etc).
The read-model is a pattern which is popular in the event-driven architecture space. There is a really good article that talks about these kinds of questions (in two parts):
https://www.infoq.com/articles/microservices-aggregates-events-cqrs-part-1-richardson
https://www.infoq.com/articles/microservices-aggregates-events-cqrs-part-2-richardson
Many common questions about microservices seem to be largely around the decomposition of a domain model, and how to overcome situations where requirements such as querying resist that decomposition. This article spells the options out clearly. Definitely worth the time to read.
In your specific case, it would mean that the File and Friends services would only need to store the primary key for the user. However, all services should publish state changes which can then be aggregated into a read-model.
If you are worry about a high volume of messages and high TPS for example 100,000 TPS for producing and consuming events I suggest that Instead of using RabbitMQ use apache Kafka or NATS (Go version because NATS has Rubby version also) in order to support a high volume of messages per second.
Also Regarding Database design you should design each micro-service base business capabilities and bounded-context according to domain driven design (DDD). so because unlike SOA it is suggested that each micro-service should has its own database then you should not be worried about normalization because you may have to repeat many structures, fields, tables and features for each microservice in order to keep them Decoupled from each other and letting them work independently to raise Availability and having scalability.
Also you can use Event sourcing + CQRS technique or Transaction Log Tailing to circumvent 2PC (2 Phase Commitment) - which is not recommended when implementing microservices - in order to exchange events between your microservices and manipulating states to have Eventual Consistency according to CAP theorem.
How do you prevent a nested attack against an Apollo server with a query such as:
{
authors {
firstName
posts {
title
author {
firstName
posts{
title
author {
firstName
posts {
title
[n author]
[n post]
}
}
}
}
}
}
}
In other words, how can you limit the number of recursions being submitted in a query? This could be a potential server vulnerability.
As of the time of writing, there isn't a built-in feature in GraphQL-JS or Apollo Server to handle this concern, but it's something that should definitely have a simple solution as GraphQL becomes more popular. This concern can be addressed with several approaches at several levels of the stack, and should also always be combined with rate limiting, so that people can't send too many queries to your server (this is a potential issue with REST as well).
I'll just list all of the different methods I can think of, and I'll try to keep this answer up to date as these solutions are implemented in various GraphQL servers. Some of them are quite simple, and some are more complex.
Query validation: In every GraphQL server, the first step to running a query is validation - this is where the server tries to determine if there are any serious errors in the query, so that we can avoid using actual server resources if we can find that there is some syntax error or invalid argument up front. GraphQL-JS comes with a selection of default rules that follow a format pretty similar to ESLint. Just like there is a rule to detect infinite cycles in fragments, one could write a validation rule to detect queries with too much nesting and reject them at the validation stage.
Query timeout: If it's not possible to detect that a query will be too resource-intensive statically (perhaps even shallow queries can be very expensive!), then we can simply add a timeout to the query execution. This has a few benefits: (1) it's a hard limit that's not too hard to reason about, and (2) this will also help with situations where one of the backends takes unreasonably long to respond. In many cases, a user of your app would prefer a missing field over waiting 10+ seconds to get a response.
Query whitelisting: This is probably the most involved method, but you could compile a list of allowed queries ahead of time, and check any incoming queries against that list. If your queries are totally static (you don't do any dynamic query generation on the client with something like Relay) this is the most reliable approach. You could use an automated tool to pull query strings out of your apps when they are deployed, so that in development you write whatever queries you want but in production only the ones you want are let through. Another benefit of this approach is that you can skip query validation entirely, since you know that all possible queries are valid already. For more benefits of static queries and whitelisting, read this post: https://dev-blog.apollodata.com/5-benefits-of-static-graphql-queries-b7fa90b0b69a
Query cost limiting: (Added in an edit) Similar to query timeouts, you can assign a cost to different operations during query execution, for example a database query, and limit the total cost the client is able to use per query. This can be combined with limiting the maximum parallelism of a single query, so that you can prevent the client from sending something that initiates thousands of parallel requests to your backend.
(1) and (2) in particular are probably something every GraphQL server should have by default, especially since many new developers might not be aware of these concerns. (3) will only work for certain kinds of apps, but might be a good choice when there are very strict performance or security requirements.
To supplement point (4) in stubailo's answer, here are some Node.js implementations that impose cost and depth bounds on incoming GraphQL documents.
graphql-depth-limit
graphql-validation-complexity
graphql-query-complexity
These are custom rules that supplement the validation phase.
A variation on query whitelisting is query signing.
During the build process, each query is cryptographically signed using a secret which is shared with the server but not bundled with the client. Then at runtime the server can validate that a query is genuine.
The advantage over whitelisting is that writing queries in the client doesn't require any changes to the server. This is especially valuable if multiple clients access the same server (e.g. web, desktop and mobile apps).
Example
In development, you write your queries as usual against your dev server which allows unsigned queries.
Then in your client build step in CI, each query is tagged with its cryptographic signature. This signature is sent by the client as a header to the server when making the request, along with the full GraphQL query string.
Your staging and production servers are configured to require a signed queries. They calculate the signature of the query received in the same way as the CI server did during the build. If the signatures don't match then they don't process the query.
Limitations:
not suitable for public facing APIs since the secret must be shared with developers
clients cannot dynamically build a GraphQL query at runtime using string interpolation, but I've never had a need for this and it is discouraged
For the Query cost limiting you could use graphql-cost-analysis
This is a validation rule which parses the query before executing it. In your GraphQL server you just have to assign a cost configuration for each field of your Schema Type Map you want.
Don't miss graphql-rate-limit 👌a GraphQL directive to add basic but granular rate limiting to your Queries or Mutations.
I want to plan a solution that manages enriched data in my architecture.
To be more clear, I have dozens of micro services.
let's say - Country, Building, Floor, Worker.
All running over a separate NoSql data store.
When I get the data from the worker service I want to present also the floor name (the worker is working on), the building name and country name.
Solution1.
Client will query all microservices.
Problem - multiple requests and making the client be aware of the structure.
I know multiple requests shouldn't bother me but I believe that returning a json describing the entity in one single call is better.
Solution 2.
Create an orchestration that retrieves the data from multiple services.
Problem - if the data (entity names, for example) is not stored in the same document in the DB it is very hard to sort and filter by these fields.
Solution 3.
Before saving the entity, e.g. worker, call all the other services and fill the relative data (Building Name, Country name).
Problem - when the building name is changed, it doesn't reflect in the worker service.
solution 4.
(This is the best one I can come up with).
Create a process that subscribes to a broker and receives all entities change.
For each entity it updates all the relavent entities.
When an entity changes, let's say building name changes, it updates all the documents that hold the building name.
Problem:
Each service has to know what can be updated.
When a trailing update happens it shouldnt update the broker again (recursive update), so this can complicate to the microservices.
solution 5.
Keeping everything normalized. Fileter and sort in ElasticSearch.
Problem: keeping normalized data in ES is too expensive performance-wise
One thing I saw Netflix do (which i like) is create intermediary services for stuff like this. So maybe a new intermediary service that can call the other services to gather all the data then create the unified output with the Country, Building, Floor, Worker.
You can even go one step further and try to come up with a scheme for providing as input which resources you want to include in the output.
So I guess this closely matches your solution 2. I notice that you mention for solution 2 that there are concerns with sorting/filtering in the DB's. I think that if you are using NoSQL then it has to be for a reason, and more often then not the reason is for performance. I think if this was done wrong then yeah you will have problems but if all the appropriate fields that are searchable are properly keyed and indexed (as #Roman Susi mentioned in his bullet points 1 and 2) then I don't see this as being a problem. Yeah this service will only be as fast as the culmination of your other services and data stores, so they have to be fast.
Now you keep your individual microservices as they are, keep the client calling one service, and encapsulate the complexity of merging the data into this new service.
This is the video that I saw this in (https://www.youtube.com/watch?v=StCrm572aEs)... its a long video but very informative.
It is hard to advice on the Solution N level, but certain problems can be avoided by the following advices:
Use globally unique identifiers for entities. For example, by assigning key values some kind of URI.
The global ids also simplify updates, because you track what has actually changed, the name or the entity. (entity has one-to-one relation with global URI)
CAP theorem says you can choose only two from CAP. Do you want a CA architecture? Or CP? Or maybe AP? This will strongly affect the way you distribute data.
For "sort and filter" there is MapReduce approach, which can distribute the load of figuring out those things.
Think carefully about the balance of normalization / denormalization. If your services operate on URIs, you can have a service which turns URIs to labels (names, descriptions, etc), but you do not need to keep the redundant information everywhere and update it. Do not do preliminary optimization, but try to keep data normalized as long as possible. This way, worker may not even need the building name but it's global id. And the microservice looks up the metadata from another microservice.
In other words, minimize the number of keys, shared between services, as part of separation of concerns.
Focus on the underlying model, not the JSON to and from. Right modelling of the data in your system(s) gains you more than saving JSON calls.
As for NoSQL, take a look at Riak database: it has adjustable CAP properties, IIRC. Even if you do not use it as such, reading it's documentation may help to come up with suitable architecture for your distributed microservices system. (Of course, this applies if you have essentially parallel system)
First of all, thanks for your question. It is similar to Main Problem Of Document DBs: how to sort collection by field from another collection? I have my own answer for that so i'll try to comment all your solutions:
Solution 1: It is good if client wants to work with Countries/Building/Floors independently. But, it does not solve problem you mentioned in Solution 2 - sorting 10k workers by building gonna be slow
Solution 2: Similar to Solution 1 if all client wants is a list enriched workers without knowing how to combine it from multiple pieces
Solution 3: As you said, unacceptable because of inconsistent data.
Solution 4: Gonna be working, most of the time. But:
Huge data duplication. If you have 20 entities, you are going to have x20 data.
Large complexity. 20 entities -> 20 different procedures to update related data
High cohesion. All your services must know each other. Data model change will propagate to every service because of update procedures
Questionable eventual consistency. It can be done so data will be consistent after failures but it is not going to be easy
Solution 5: Kind of answer :-)
But - you do not want everything. Keep separated services that serve separated entities and build other services on top of them.
If client wants enriched data - build service that returns enriched data, as in Solution 2.
If client wants to display list of enriched data with filtering and sorting - build a service that provides enriched data with filtering and sorting capability! Likely, implementation of such service will contain ES instance that contains cached and indexed data from lower-level services. Point here is that ES does not have to contain everything or be shared between every service - it is up to you to decide better balance between performance and infrastructure resources.
This is a case where Linked Data can help you.
Basically the Floor attribute for the worker would be an URI (a link) to the floor itself. And Any other linked data should be expressed as URIs as well.
Modeled with some JSON-LD it would look like this:
worker = {
'#id': '/workers/87373',
name: 'John',
floor: {
'#id': '/floors/123'
}
}
floor = {
'#id': '/floor/123',
'level': 12,
building: { '#id': '/buildings/87' }
}
building = {
'#id': '/buildings/87',
name: 'John's home',
city: { '#id': '/cities/908' }
}
This way all the client has to do is append the BASE URL (like api.example.com) to the #id and make a simple GET call.
To remove the extra calls burden from the client (in case it's a slow mobile device), we use the gateway pattern with micro-services. The gateway can expand those links with very little effort and augment the return object. It can also do multiple calls in parallel.
So the gateway will make a GET /floor/123 call and replace the floor object on the worker with the reply.