I have a node type that has a string property that will have the same value really often. Etc. Millions of nodes with only 5 options of that string value. I will be doing searches by that property.
My question would be what is better in terms of performance and memory:
a) Implement it as a node property and have lots of duplicates (and search using WHERE).
b) Implement it as 5 additional nodes, where all original nodes reference one of them (and search using additional MATCH).
Without knowing further details it's hard to give a general purpose answer.
From a performance perspective it's better to limit the search as early as possible. Even more beneficial if you do not have to look into properties for a traversal.
Given that I assume it's better to move the lookup property into a seperate node and use the value as relationship type.
Use labels; this blog post is a good intro to this new Neo4j 2.0 feature:
Labels and Schema Indexes in Neo4j
I've thought about this problem a little as well. In my case, I had to represent state:
STARTED
IN_PROGRESS
SUBMITTED
COMPLETED
Overall the Node + Relationship approach looks more appealing in that only a single relationship reference needs to be maintained each time rather than a property string and you don't need to scan an extra additional index which has to be maintained on the property (memory and performance would intuitively be in favor of this approach).
Another advantage is that it easily supports the ability of a node being linked to multiple "special nodes". If you foresee a situation where this should be possible in your model, this is better than having to use a property array (and searching using "in").
In practice I found that the problem then became, how do you access these special nodes each time. Either you maintain some sort of constants reference where you have the node ID of these special nodes where you can jump right into them in your START statement (this is what we do) or you need to do a search against property of the special node each time (name, perhaps) and then traverse down it's relationships. This doesn't make for the prettiest of cypher queries.
Related
I have an entity that represents a relationship between two entity groups but the entity belongs to one of the groups. However, my queries for this data are going to be mostly with the other entity group. To support the queries I see I have two choices a) Create a global index that has the other entity group key as prefix b) Move the entity into the other entity group and create ancestor index.
I saw a presentation which mentioned that ancestor indexes map internally to a separate table per entity group while there is a single table for the global index. That makes me feel that ancestors are better than using global indexes which includes the ancestor keys as prefix for this specific use case where I will always be querying in the context of some ancestor key.
Looking for guidance on this in terms of performance, storage characteristics, transaction latency and any other architectural considerations to make the final call.
From what I was able to found I would say it depends on the of the type of work you'll be doing. looked at this docs and it suggest you Avoid writing to an entity group more than once per second. Also indexing a property could result in increased latency. Also it states that If you need strong consistency for your queries, use an ancestor query, in that docs there are many advice's on how to avoid latency and other issues. it should help you on taking a call.
I ended up using a 3rd option which is to have another entity denormalized into the other entity group and have ancestor queries on it. This allows me to efficiently query data for either of the entity groups. Since I was using transactions already, denormalizing wouldn't cause any inconsistencies and everything seems to work well.
My table is (device, type, value, timestamp), where (device,type,timestamp) makes a unique combination ( a candidate for composite key in non-DynamoDB DBMS).
My queries can range between any of these three attributes, such as
GET (value)s from (device) with (type) having (timestamp) greater than <some-timestamp>
I'm using dynamoosejs/dynamoose. And from most of the searches, I believe I'm supposed to use a combination of the three fields (as a single field ; device-type-timestamp) as id. However the set: function of Schema doesn't let me use the object properties (such as this.device) and due to some reasons, I cannot do it externally.
The closest I got (id:uuidv4:hashKey, device:string:GlobalSecIndex, type:string:LocalSecIndex, timestamp:Date:LocalSecIndex)
and
(id:uuidv4:rangeKey, device:string:hashKey, type:string:LocalSecIndex, timestamp:Date:LocalSecIndex)
and so on..
However, while using a Query, it becomes difficult to fetch results of particular device,type as the id, (hashKey or rangeKey) keeps missing from the scene.
So the question. How would you do it for such kind of table?
And point to be noted, this table is meant to gather content from IoT devices, which is generated every 5 mins by each device on an average.
I'm curious why you are choosing DynamoDB for this task. Advanced queries like this seem to be much better suited for a SQL based database as opposed to a NoSQL database. Due to the advanced nature of SQL queries, this task in my experience is a lot easier in SQL databases. So I would encourage you to think about if DynamoDB is truly the right system for what you are trying to do here.
If you determine it is, you might have to restructure your data a little bit. You could do something like having a property that is device-type and that will be the device and type values combined. Then set that as an index, and query based on that and sort by the timestamp, and filter out the results that are not greater than the value you want.
You are correct that currently, Dynamoose does not pass in the entire object into the set function. This is something that personally I'm open to exploring. I'm a member on the GitHub project, and if you would like to submit a PR adding that feature I would be more than happy to help explore that option with you and get that into the codebase.
The other thing you might want to explore is having a DynamoDB stream, that will set that device-type property whenever it gets added to your DynamoDB table. That would abstract that logic out of DynamoDB and your application. I'm not sure if it's necessary for what you are doing to decouple it to that level, but it might be something you want to explore.
Finally, depending on your setup, you could figure out which item will be more unique, device or type, and setup an index on that property. Then just query based on that, and filter out the results of the other property that you don't want. I'm not sure if that is what you are looking for, it will of course work, but I'm not sure how many items you will have in your table, and there get to be questions about scalability at a certain level. One way to solve some of those scalability questions might be to set the TTL of your items if you know that you the timestamp you are querying for is constant, or predictable ahead of time.
Overall there are a lot of ways to achieve what you are looking to do. Without more detail about how many items, what exactly those properties will be doing, the amount of scalability you require, which of those properties will be most unique, etc. it's hard to give a good solution. I would highly encourage you to think about if NoSQL is truly the best way to go. That query you are looking to do seems a LOT more like a SQL query. Not saying it's impossible in DynamoDB, but it will require some thought about how you want to structure your data model, and such.
Considering opinion of #charlie-fish, I decided to jump into Dynamoose and improvise the code to pass the model to the set function of the attribute. However, I discovered that the model is already being passed to default parameter of the attribute. So I changed my Schema to the following:
id:hashKey;default: function(model){ return model.device + "" + model.type; }
timestamp:rangeKey
For anyone landing here on this answer, please note that the default & set functions can access attribute options & schema instance using this . However both those functions should be regular functions, rather than arrow functions.
Keeping this here as an answer, but I won't accept it as an answer to my question for sometime, as I want to wait for someone else to hit out a better approach.
I also want to make sure that if a value is passed for id field, it shouldn't be set. For this I can use set to ignore the actual incoming value, which I don't know how, as of yet.
Official recommendation from the team is, to my knowledge, to put all datatypes into single collection that have something like type=someType field on documents to distinguish types.
Now, if we assume large databases with partitioning where different object types can be:
Completely different fields (so no common field for partitioning)
Related (through reference)
How to organize things so that things that should go together end up in same partition?
For example, lets say we have:
User
BlogPost
BlogPostComment
If we store them as separate types with type=user|blogPost|blogPostComment, in same collection, how do we ensure that user, his blogposts and all the corresponding comments end up in same partition?
Is there some best practice for this?
[UPDATE]
Can you ever avoid cross-partition queries completely? Should that be a goal? Or you just try to minimize them?
For example, you can partition your data perfectly for 99% of cases/queries but then you need some dashboard to show aggregates from all-the-data. Is that something you just accept as inevitable and try to minimize or is it possible to avoid it completely?
I've written about this somewhat extensively in other similar questions regarding Cosmos.
Basically, when dealing with many different logical entity types in a single Cosmos collection the easiest option is to put a generic (or abstract, as you refer to it) partition key on all your documents. At this point it's the concern of the application to make sure that at runtime the appropriate value is chosen. I usually name this document property either partitionKey, routingKey or something similar.
This is extremely important when designing for optimal query efficiency as your choice of partition keys can have a huge impact on query and throughput performance. A generic key like this lets you design the optimal storage of your data as it benefits whatever application you're building.
Even something like tenant does not make sense as different tenants might have wildly different data size and access patterns. Instead you could include the tenantId at runtime as part of your partition key as a kind of composite.
UPDATE:
For certain query patterns it might be possible to serve them entirely out of a single partition. It's definitely not the end of the world if things end up going cross partition though. The system is still quick. If possible, limiting the amount of partitions that need to be touched for a given query is ideal but you're never going to get away from it 100% of the time.
A partition should hold data related to a group that is expected to grow, for instance a Tenant which will group many documents (which can be of different types as you have mentioned) So the Partition Key in this instance should be the TenantId. The partitioning is more about the data relating to a group than the type of data. If the data is related to a User then you could use the UserId, however many users may comment on the same posts so it doesn't seem like a good candidate for a partition key unless there is some de-normalization of the user info so it doest have to relate back to the other users directly.. if that makes sense?
I want perform a weighted search in cts:collection-query. Is there any way provided for this?
What exactly I want to do is I want to fetch documents from a collection and give them different weight in a similar way as we do in cts:element-range-query.
cts:collection-query does not have any scoring options, unlike cts:element-range-query. A document either matches a collection query or it doesn't.
One option for you is to move the information you're current modeling with collections into elements (or JSON properties) within the documents; then you'll be able to use cts:element-range-query.
You haven't specified what kind of information you're using the collections for; it's hard to picture typical collection names benefitting from this approach. Some more detail might make that more clear.
If documents in some collections are "better" (should score higher) than ones in other collections, and those valuations are static, you could set the document quality based on the collections it belongs to. Not exactly the same, but perhaps that accomplishes the goal.
While using Graph Databases(my case Neo4j), we can represent the same information many ways. Making each entity a Node and connecting all entities through relationships or just adding the entities to attribute list of a Node.diff
Following are two different representations of the same data.
Overall, which mechanism is suitable in which conditions?
My use case involves traversing the Database from different nodes until 4 depths and examining the information through connected nodes or attributes (based on which approach it is).
One query of interest may be, "Who are the friends of John who went to Stanford?"
What is the difference in terms of Storage, computations
Normally,
properties are loaded lazily, and are more expensive to hold in cache, especially strings. Nodes and Relationships are most effective for traversal, especially since the relationships types are stored together with the relatoinship records and thus don't trigger property loads when used in traversals.
Also, a balanced graph (that is, not many dense nodes with over say 10K relationships) is most effective to traverse.
I would try to model most of the reoccurring proeprties as nodes connecting to the entities, thus using the graph itself to index on these values, instead of having to revert to filter on property values or index the property with an expensive index lookup.
The first one is much better since you're querying on entities such as Stanford- and that entity is related to many person nodes. My opinion that modeling as nodes is more intuitive and easier to query on. "Find all persons who went to Stanford" would not be very easy to do in your second model as you don't have a place to start traversing from.
I'd use attributes mainly to describe the node/entity use them to filter results from the query e.g. Who are friends of John who went to Stanford in the year 2010. In this case, the year attribute would just be used to trim the results. Depends on your use case- if year is really important and drives a lot of queries or is used to represent a timeline, you could even model the year as a node attached to Stanford.