You are an Enterprise administrator for contoso.com. The corporate network of the company consists of a single Active Directory domain. You have been asked to deploy file servers that run Windows Server 2008 and ensure that the file server support volumes larger than 2 terabytes.
You also need to ensure that if a single server fails, access to all data is maintained and if a single disk fails, the data redundancy is maintained. You also need to maximize the disk throughput
Which of the following options would you choose to accomplish the assigned task? (Select 2. Each correct answer will present a part of the solution)
A.
Deploy a Windows Server 2008 server and connect an external storage subsystem to it that supports Microsoft Multipath I/O.
B.
Deploy a two-node failover cluster. Connect an external storage subsystem.
C.
Configure the external storage subsystem as a RAID 1 array and format the array as an MBR disk.
D.
Configure the external storage subsystem as a RAID 10 array and format the array as a GPT disk.
Explanation:
To ensure that if a single server fails, access to all data is maintained and if a single disk fails, the data redundancy is maintained, you need to deploy a two-node failover cluster. Connect an external storage subsystem. Configure the external storage subsystem as a RAID 10 array. Format the array as a GPT disk.
A combining the different RAID levels gives us the option of RAID10. RAID10 is equivalent toRAID1 + 0. So, you can have a few disks (at least 4 and always even numbers) and mirror the drives two at a time. This gives the redundancy. Then you take those mirrors and combine them into a RAID 0 stripe. This allows redundancy, faster read operations, and fast writes (avoiding a parity calculation). RAID1 is a mirror which is faster than a single disk, but not as fast for read operations as 3+ disks (RAID1 is just 2 disks). RAID5 is a stripe with parity which is faster on read operations than RAID1 but not ideal for write operations because it is required to calculate a parity block of data.
Reference: Brad Kingsley’s Blog
http://blogs.orcsweb.com/brad/archive/2007/08/06/raid10.aspx