From a transaction standpoint, there are ways to optimize, such as writing multiple entities within a given partition at once, which can reduce your transaction rate depending on how many you can batch together in a single write.
Also, consider using a map/reduce implementation. This will be an exercise for you to explore further, as we really don't know your schema or viability of map/reduce with your data. That said, it could save you transactions.
Next: SQL Azure, at 40GB, would run about $400 / month. Your solution might fit better with a relational store, depending on searching needs, and pricing isn't that much different (and with no transaction costs).
Finally: Instead of focusing on your storage and transactional cost alone, look at the big picture, and what you're getting (durable, replicated storage; ability to scale to 100TB per account, etc.). And think about your business model, and what it would cost to purchase your own servers (or what it would cost for dedicated hosting). As others have already stated, it's unlikely you'll get the performance you're looking for in a traditional hosting offering, since the low-price systems are shared systems with no SLA.
if you buy a traditional hosting server you'll spend 50 USD per month
I'm not sure a webhosting company would let you do 10 million transactions a day at 50 USD, really. Anyway if one is crazy to say yes at this price, why don't you go for it and forget about Azure ?
300 USD per month for 10 million transactions a day sounds totally cheap, David is right.
发布评论
评论（2）
一些想法：
从事务的角度来看，有一些方法可以优化，例如一次在给定分区中写入多个实体，这可以降低事务率，具体取决于一次写入中可以批量处理的实体数量。
另外，考虑使用映射/归约实现。这将是您进一步探索的练习，因为我们确实不知道您的架构或映射/归约与您的数据的可行性。也就是说，它可以节省您的交易。
接下来：SQL Azure，40GB，每月运行费用约为 400 美元。您的解决方案可能更适合关系商店，具体取决于搜索需求，并且定价差别不大（并且没有交易成本）。
最后：不要只关注存储和交易成本，而是要着眼于大局以及您所获得的内容（持久的复制存储；每个帐户扩展到 100TB 的能力等）。并考虑您的业务模式，以及购买自己的服务器的成本（或专用托管的成本）。正如其他人已经指出的那样，您不太可能在传统托管产品中获得所需的性能，因为低价系统是没有 SLA 的共享系统。
A few thoughts:
From a transaction standpoint, there are ways to optimize, such as writing multiple entities within a given partition at once, which can reduce your transaction rate depending on how many you can batch together in a single write.
Also, consider using a map/reduce implementation. This will be an exercise for you to explore further, as we really don't know your schema or viability of map/reduce with your data. That said, it could save you transactions.
Next: SQL Azure, at 40GB, would run about $400 / month. Your solution might fit better with a relational store, depending on searching needs, and pricing isn't that much different (and with no transaction costs).
Finally: Instead of focusing on your storage and transactional cost alone, look at the big picture, and what you're getting (durable, replicated storage; ability to scale to 100TB per account, etc.). And think about your business model, and what it would cost to purchase your own servers (or what it would cost for dedicated hosting). As others have already stated, it's unlikely you'll get the performance you're looking for in a traditional hosting offering, since the low-price systems are shared systems with no SLA.
我不确定网络托管公司是否会让您每天以 50 美元的价格进行 1000 万笔交易，真的。
无论如何，如果有人疯狂地以这个价格答应，为什么不去买它并忘记 Azure 呢？
每月 300 美元，每天 1000 万笔交易听起来非常便宜，大卫是对的。
I'm not sure a webhosting company would let you do 10 million transactions a day at 50 USD, really.
Anyway if one is crazy to say yes at this price, why don't you go for it and forget about Azure ?
300 USD per month for 10 million transactions a day sounds totally cheap, David is right.